-
Notifications
You must be signed in to change notification settings - Fork 0
/
db_sample.txt
119 lines (119 loc) · 111 KB
/
db_sample.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
자율주행을 위한 융합기술을 이용한 비전 기반의 관심영역 설정 장치 본 발명은 자율주행을 위한 융합기술을 이용한 비전 기반의 관심영역 설정 장치에 관한 것으로, 레이더 장치가 감지한 물체의 위도와 감도 정보를 포함하는 수신감도를 제공하는 좌표정보 획득모듈; 레이더로부터 제공하는 수신감도를 수신하고, 수신감도에 포함된 물체의 위치정보를 이미지상에서 물체의 관심위치(POI)로 변환하여 비전모듈에 제공하는 변환모듈; 및 변환모듈로부터 수신되는 물체의 관심위치(POI)를 기반으로 카메라를 통해 획득하는 이미지에서 물체의 관심영역(ROI)을 검출하는 비전모듈;을 포함한다.
비전센서를 이용한 자율주행차량의 교차로 통과지원시스템 및 통과지원방법 본 발명은 비전센서를 이용한 자율주행차량의 교차로 통과지원시스템 및 통과지원방법에 대한 것이다. 보다 상세하게는 교차로 통과지원 시스템에 있어서, 교차로 주변 일측에 설치되어, 실시간으로 접근하는 접근차량의 영상을 획득하여 상기 영상을 기반으로 접근차량 신호등의 현재상태를 기반으로 위험도를 계산하는 교차로 통과지원장치; 및 자율주행차량에 구비되어, 상기 교차로 통과지원장치에서 전송받은 위험도를 기반으로 이동여부를 판단하여 상기 자율주행차량의 주행을 제어하는 자율주행차량제어기;를 포함하는 것을 특징으로 하는 비전센서를 이용한 자율주행차량 교차로 통과지원시스템에 관한 것이다.
자율 주행을 위한 비전-기반 샘플-효율적 강화 학습 프레임워크 프레임워크는 자율 주행을 위한 안내된 정책 탐색에 기초하여 비전 및 샘플-효율적 강화-학습을 조합한다. 컨트롤러는 비전으로부터 환경 정보를 추출하고 강화 학습을 사용하여 주행하도록 트레이닝된다.
자율 주행 차량을 위한 비전 센서의 제어 방법 및 장치 비전 센서의 제어 방법 및 장치가 개시된다. 일 실시예에 따른 비전 센서의 제어 장치는 호스트 차량의 주행 경로 상 임계값 이상의 조도 변화가 예상되는 지점을 예측하고, 호스트 차량이 조도 변화가 예상되는 지점으부터 임계 거리 내에 위치하는 경우 예상되는 조도 변화에 기초하여 비전 센서를 제어한다.
페일세이프 기능을 구비한 비전 시스템 및 그 비전 시스템의 페일세이프 방법 본 발명은 비전 시스템의 사용이 가능한 경우 카메라로부터 획득된 영상정보에 근거하여 무인 자율 주행이 가능하도록 하고, 비전 시스템의 사용이 불가능한 경우 차량에 설치된 센서로부터 수신된 감지신호에 의해 차량의 자율 주행이 가능하도록 한 페일세이프 기능을 구비한 비전 시스템 및 그 비전 시스템의 페일세이프 방법에 관한 것이다. 이를 위하여 본 발명은 빛의 양을 수신하는 단계; 상기 수신된 빛의 양과 미리 설정된 임계값과의 비교를 통해 비전 시스템의 사용 가능 여부를 결정하는 단계; 및 상기 비전 시스템의 사용이 불가능한 경우 상기 차량에 설치된 대체부로부터 수신된 감시신호에 근거하여 상기 차량이 자율적으로 주행되도록 제어하는 단계를 포함하는 것을 특징으로 하는 비전 시스템의 페일세이프 방법을 제공한다.
스테레오 카메라의 출력영상을 기반으로 한 자율주행 자동차의 차선 인식 방법 본 발명은 자율주행 자동차에서 스테레오 카메라의 출력영상을 기반으로 차선을 인식하는 기술에 관한 것이다. 이러한 본 발명은 도로 영상을 획득하는 단계; 도로면 인식 및 분리단계; 및 인식하는 차선 인식단계;를 포함하되, 상기 차선 인식단계는 실시간 양안 스테레오카메라를 이용하여 입력 영상으로부터 도로면 영상을 분리하고, 분리된 도로면 영상에 다이나믹프로그래밍과 칼만필터를 적용하여 안정적으로 차선을 인식할 수 있도록 하는 것을 특징으로 한다.
가변 차로 인식 및 이용 여부 판단 시스템 및 방법 본 발명은 가변 차로 인식 및 이용 여부 판단 시스템 및 방법에 관한 것으로서, 차량 동역학 정보를 감지하는 차량 동역학 센서, 차량 전방 영상을 인식하는 차량 전방위 감시 카메라, 차량 주변 정보를 감지하는 차량 주변 감시 센서 및 차량의 현재 위치, 주행 경로 및 교통량을 제공하는 내비게이션을 포함하는 감지 장치들 및 상기 감지 장치들로부터 감지 신호들을 입력 받아 차량 전방의 가변 차로를 인식하고, 인식된 상기 가변 차로의 사용 가부를 판단하며 그에 따라 상기 가변 차로의 이용 여부를 결정하는 전자 제어 장치를포함함으로써, 도로 교통을 보다 원활하게 할 수 있다.
자율주행 무인차량에 탑재된 쌍안카메라를 이용한 동적물체 검출방법 본 발명은 자율주행 무인차량에 서로 이격되게 설치된 카메라를 이용하여, 전방의 동적물체를 감지하도록 한 자율주행 무인차량에 탑재된 쌍안카메라를 이용한 동적물체 검출방법에 관한 것이다. 본 발명에 따른 자율주행 무인차량에 탑재된 쌍안카메라를 이용한 동적물체 검출방법은, 주행중인 자율주행 무인차량의 전방에 위치한 동적물체를 탐지하는 방법에 있어서, 주행중인 자율주행 무인차량의 주행정보와, 상기 자율주행 무인차량에 탑재된 쌍안 카메라에서 촬영한 이미지가 입력되는 데이터 입력단계(S110)와, 상기 쌍안 카메라로부터 동시에 촬영된 한 쌍의 이미지로부터 스테레오 정합을 실시하여 특징점 추출하는 스테레오 정합단계(S120)와, 상기 이미지에서 정해진 구간의 값을 갖는 영역이 하나의 수퍼픽셀(super pixel)이 되도록 상기 이미지를 복수의 수퍼픽셀로 분할하는 수퍼픽셀 변환단계(S130)와, 상기 복수의 수퍼픽셀마다 고차원 마르코프 랜덤필드를 계산하여 동적물체가 존재하는 수퍼픽셀을 검출하는 고차원 마르코프 랜덤필드 계산단계(S140)와, 상기 고차원 마르코프 랜덤필드 계산단계(S140)에 동적물체가 있는 수퍼픽셀에 동적물체가 있는 것으로 판단하는 최적정보선택단계(S150)를 포함한다.
카메라 및 라이다 센서를 이용한 차량 자율주행방법 본 발명에서는 카메라 및 라이다 센서를 이용한 차량 자율주행 방법에 관한 것으로, 차량의 전방에 설치된 좌측 라이다 센서와 우측 라이다 센서에서 센서값을 획득하고, 상기 좌측 라이다 센서 또는 우측 라이다 센서가 물체를 감지하고 하나 또는 두 개의 라이다 센서에서 감지된 물체와의 거리를 확인하여 비상여부를 결정하며, 물체와의 거리를 확인하여 설정값 미만일 경우 비상정지를 진행하도록 신호를 전송하며; 비상정지를 진행할 필요가 없는 경우, 회전 라이다 센서를 통하여 차량 주변에 장애물이 있는지 확인하고, 획득한 회전 라이다 센서값을 통하여 물체의 위치를 인식하고 장애물 회피를 결정하며, 차량 주변에 물체가 인식되면 장애물 회피를 진행하도록 신호를 전송하며; 차량 주변에 장애물이 없을 경우 웹캠 카메라를 통하여 데이터를 획득하며; 획득된 데이터를 통하여 차선을 감지하고, 감지된 차선을 따라 차선을 추적하며; 비상정지 신호를 최우선으로 속도 및 조향 명령을 전송하며, 다음 우선순위로 장애물 회피의 신호로 속도 및 조향 명령을 전송하며, 다음으로 차선 추적을 통한 속도 및 조향 명령을 전송하여 주행하는 것을 포함하는 카메라 및 라이다 센서를 이용한 차량 자율주행방법을 제안한다.
자율주행용 서라운드 카메라 시스템 본 발명은 차량 주변 360도 전체를 실시간으로 감시함은 물론, 안개, 비, 눈, 황사, 무광(야간), 역광 등과 같은 환경하에서도 차량 주행에 방해되는 대상물을 감지하여 운전자에게 경고 및 가이드함은 물론, 자율주행차량에도 적용하여 운전자의 안전을 확보할 수 있는 자율주행용 서라운드 카메라 시스템에 관한 것으로서, 더욱 상세하게는, 차량의 전방을 감시하는 적어도 하나의 전방감시카메라; 차량의 후방을 감시하는 적어도 하나의 후방감시카메라; 차량의 양 측방을 감시하는 적어도 하나의 측방감시카메라; 전방감시카메라, 후방감시카메라 및 측방감시카메라를 통해 취득된 각각의 영상 또는 각각의 영상으로부터 둘 이상을 조합한 영상 또는 각각의 영상 중 적어도 어느 하나의 영상과 조합한 영상으로 구성된 듀얼 영상을 디스플레이하는 모니터링수단; 및 전방감시카메라, 후방감시카메라 및 측방감시카메라로부터 취득된 영상을 처리하여 모니터링수단으로 전송하고, 취득된 영상으로부터 차량의 주행에 방해되는 대상물을 감지하고 분석하여 운전자에게 경고하며, 차량의 전자제어유닛(ECU)을 컨트롤하는 비디오제어유닛(VCU : Video Control unit);을 포함하고, 전방감시카메라, 후방감시카메라 및 측방감시카메라를 통해 차량 주변 360도 전방위를 감시하는 것을 특징으로 하는 자율주행용 서라운드 카메라 시스템에 관한 것이다.
카메라와 그 제어 방법 및 카메라를 포함한 자율 주행 시스템 카메라와 그 제어 방법 및 카메라를 포함한 자율 주행 시스템이 개시된다. 본 발명의 카메라는 전기적으로 제어되어 적외선 파장의 빛을 통과 및 차단하는 액티브 필터; 상기 액티브 필터를 통과하는 빛을 전기적인 신호로 변환하여 영상 신호를 출력하는 이미지 센서; 상기 이미지 센서로부터 획득된 영상 신호를 분석하는 영상 분석부; 및 상기 영상 신호의 분석 결과를 바탕으로 상기 액티브 필터의 동작 모드를 전기적으로 제어하여 상기 액티브 필터를 통과하는 빛의 파장을 선택하는 필터 제어부를 포함할 수 있다.
스테레오 카메라 기반의 자율 주행 방법 및 그 장치 스테레오 카메라 기반의 자율 주행 방법 및 그 장치가 개시된다. 일 실시예에 따른 자율 주행 방법은, 차량의 주행 상황에 기초하여 스테레오 카메라의 폭과 관련된 파라미터를 결정하는 단계와, 촬영부에 포함된 두 대의 카메라들이 상기 파라미터에 대응하는 폭을 가지도록 촬영부를 제어하는 단계들을 포함한다.
카메라 영상정보와 차량정보를 융합하여 전후방 차선을 인지하는 차선 감지 방법 및 장치 본 발명은 카메라 영상정보와 자차 속도/헤딩 정보 등 차량정보 융합 방식에 따라 합류점이나 분기점과 같이 도로가 특성이 갑자기 변하는 경우 등에 있어서 종래 보다 차선을 더 정확하게 추정할 수 있으며, 전방/후방 차선을 더 정확하게 인지할 수 있기 때문에 차후 차선 기반 자율 주행 시스템에서 차선 유지 자율주행, 차선 변경 자율 주행 등을 위하여 위험도 판단 등에서 좀 더 정확한 처리가 가능하게 하는 차선 감지 방법 및 장치에 관한 것이다.
차량 위치 보정 시스템 및 방법, 그리고 카메라 본 개시는 차량 위치 보정 시스템 및 방법, 그리고 카메라에 관한 것이다. 구체적으로, 본 개시의 차량 위치 보정 시스템은, 차량의 전방 및 후방에 대한 이미지 데이터에 기초하여 차량의 주행차로의 전방 차선 및 후방 차선을 검출하고, 차량의 주행차로의 전방 차선 및 후방 차선에 기초하여 전방 차선에 대한 전방 차선 오프셋과 후방 차선에 대한 후방 차선 오프셋을 계산하고, 전방 차선 오프셋과 후방 차선 오프셋을 비교하여 상기 비교의 결과에 따라 차량의 최종 차선 오프셋을 결정한다.
차량용 영상 처리 장치 및 방법 본 발명은 차량용 블랙박스에 기본 장착되어 있는 카메라의 영상을 직접 해석함으로써 도플러 효과가 거의 없어 분석 단계를 간소화하고, 급제동 및 사고 발생 이전에 위험을 감지할 수 있는 수단을 포함하고 있는 차량용 블랙박스 시스템의 영상 처리 장치 및 방법에 관한 것이다. 이를 위한 본 발명은, 카메라에 의해 촬영된 영상 내에 있는 피사체의 크기 변화를 분석하여 피사체와의 거리 변화를 감지하는 피사체 거리 변화 감지부; 상기 영상 내에 있는 광원을 분석하는 광원 분석부; 상기 영상을 구역별로 분할하여 분할된 구역별로 사고 위험도의 가중치를 다르게 하기 위한 영상 구역 분할부; 상기 영상 구역 분할부에 존재하는 사고 위험 상항에 대응되는 경보를 발생하는 경보부;를 포함하는 것을 특징으로 한다.
자율주행을 위한 학습대상 이미지 추출 장치 및 방법 자율주행을 위한 학습대상 이미지 추출 방법 및 장치가 개시된다. 상기 방법은, 원시 도로주행 영상을 수신하는 원시 동영상 수신 단계, 상기 원시 도로주행 영상의 메타데이터에 기초하여 추출 개수를 결정하고, 상기 원시 도로주행 영상으로부터 상기 결정된 개수의 프레임 이미지를 추출하는 1차 샘플링 단계, 및 상기 결정된 개수의 프레임 이미지 각각의 제1 학습 대상 객체의 포함 여부 및 상기 제1 학습 대상 객체의 이동량에 기초하여, 상기 결정된 개수의 프레임 이미지로부터 하나 이상의 학습 대상 프레임 이미지를 추출하는 2차 샘플링 단계를 포함하고, 상기 원시 도로주행 영상의 상기 메타데이터는 상기 원시 도로주행 영상과 관련된 상기 차량의 GPS 정보, 상기 차량의 이동 속도 정보 및 상기 원시 도로주행 영상의 촬영 시간 중 하나 이상을 포함한다.
자율주행을 위한 학습대상 이미지 패키징 장치 및 방법 자율주행을 위한 학습대상 이미지 패키징 장치 및 방법이 제공된다. 상기 방법은 원시 도로주행 영상을 수신하는 원시 동영상 수신 단계, 상기 원시 도로주행 영상과 관련된 지역의 밀집변수에 기초하여 추출 개수를 결정하고, 상기 원시 도로주행 영상으로부터 상기 결정된 개수의 프레임 이미지를 추출하는 1차 샘플링 단계, 및 상기 결정된 개수의 프레임 이미지 각각의 제1 학습 대상 객체의 포함 여부 및 상기 제1 학습 대상 객체의 이동량에 기초하여, 상기 결정된 개수의 프레임 이미지로부터 하나 이상의 학습 대상 프레임 이미지를 추출하는 2차 샘플링 단계를 포함하고, 상기 원시 도로주행 영상과 관련된 지역의 밀집변수는 상기 원시 도로주행 영상과 관련된 지역의 인구 통계 데이터 또는 차량 통계 데이터 중 하나 이상을 포함한다.
AGV 주행제어를 위한 인공신경망 학습 방법 및 장치 본 발명은 AGV 주행제어를 위한 인공신경망 학습 방법 및 장치에 관한 것으로서, 더욱 상세하게는 카메라를 통하여 수집한 영상 데이터로부터, 머신러닝(machine learning)에 의해 물체 인식을 위한 인공신경망 모델을 형성 및 업데이트하여 이를 AGV에 장착된 AGV 주행제어 장치에 제공하는, AGV 주행제어를 위한 인공신경망 학습 방법 및 장치에 관한 것이다. 본 발명에 의하면, 소량의 데이터 학습만으로 이미지 인식이 가능한 “퓨샷러닝” 기술을 적용하되, label된 소량의 데이터를 사용하여 label되지 않은 데이터에 대한 학습 정확도를 높이는 반지도 학습(semi-supervised learning)으로 퓨샷러닝의 정확도를 높여, AGV의 자율주행을 위한 인공신경망 모델을 학습시키고 형성 및 업데이트하는 방법 및 장치를 제공하며, 또한 또한 AGV가 아닌 별도의 장치에서 지속적인 인공신경망 모델의 학습에 의하여 업데이트된 인공신경망 모델을 생성하고 이를 AGV에 제공하여 줌으로써, AGV는 직접 인공신경망 모델의 업데이트 생성에 의한 부담없이, 항상 최신의 인공신경망 모델에 의해 주행제어를 할 수 있게 되도록 한다.
주행상황에서의 동시적 물체 인식 및 학습 방법 본 발명은, 주행상황에서의 동시적 물체 인식 및 학습 방법에 관한 것으로서, 특히 영상신호로부터 오브젝트의 검출을 수행하면서 동시에 자동적인 학습을 수행하여 주행상황에서 검출기의 성능을 지속적으로 업그레이드 하는 동시적 물체 인식 및 학습 방법에 관한 것이다. 본 발명은, 주행상황에서 카메라에 의해 촬영된 영상신호로부터 오브젝트를 인식하는 물체 인식 및 학습 방법에 있어서, 영상처리부가 영상신호에 RPN(Region Proposal Network)기법을 적용하여 오브젝트를 판단할 수 있는 제1 오브젝트 후보 영역을 추출하고 기 저장된 영상신호로부터 오브젝트를 트랙킹하여 제2 오브젝트 후보 영역을 추출하는 제1 단계; 검출부가 제1 및 제2 오브젝트 후보 영역으로부터 오브젝트를 추정하고 실제 물체의 종류와 비교하여 일치도를 나타내는 확률값을 추정 및 저장하는 제2 단계; 및 백워딩부에서 검출부에서 추정된 오브젝트 및 확률값을 물체의 종류마다 기 설정된 임계값과 비교하여 임계값 이상일 경우 오브젝트의 종류 및 확률값을 산출하고 영상처리부로 오브젝트의 종류 및 확률값을 피드백하여 물체와 관련된 제1 및 제2 오브젝트 후보 영역을 업데이트 하는 제3 단계;를 포함하는 주행상황에서의 동시적 물체 인식 및 학습 방법을 제공한다.
의료용 카메라의 USB2.0 영상처리장치 본 고안은 의료용 카메라로부터 영상데이터를 입력받는 영상입수부, 입수된 아날로그 영상을 디지털 데이터로 변환하는 영상처리부, 디지털 데이터를 전송하는 USB2.0 송수신부, 전원장치를 포함하는 영상처리 장치가 제공된다. 영상입수부는 아날로그 비디오 표준인 Composite 혹은 SVHS 단자를 통하여 영상데이터를 입수 받게 된다. 대부분의 카메라는 영상데이터 출력을 Composite 혹은 SVHS로 출력하게 되므로 영상을 입수 받기가 용이하게 된다. 입수된 영상데이터는 AD Converter를 통하여 디지털 시그널로 변환되고, 변환된 시그널을 USB2.0 송수신부를 통하여 컴퓨터에 전송하게 된다. 이때 전송되는 속도는 USB 1.1에 비하여 약 40배 정도의 속도로 전송되므로 압축하지 않고 동영상을 30fr/sec속도로 컴퓨터에 전송하는 것이 가능하게 된다.
의료 영상의 메타데이터 예측 장치 및 방법 본 개시는 신경망을 이용하여 의료 영상 및 메타데이터의 관계를 기계학습하고, 의료영상에 기계학습모델을 적용하여 메타데이터를 획득하는 장치 및 방법에 관한 것으로서, 학습(training)을 위한 복수의 의료 영상 및 복수의 의료 영상 각각에 매칭된 메타데이터를 기초로, 의료 영상의 메타데이터를 예측하는 예측 모델을 학습하는 단계, 및 학습된 예측 모델을 이용하여, 입력된 의료 영상에 대한 메타데이터를 예측하는 단계를 포함한다
의료 영상 신호 세기 정규화를 통한 관심 부위 검출 방법 및 시스템 의료 영상 신호 세기 정규화를 통한 관심 부위 검출 방법 및 시스템이 제공된다. 본 발명의 실시예에 따른 관심 부위 검출 방법은, 의료 영상을 구성하는 픽셀들의 신호 세기들을 산출하고, 의료 영상의 신호 세기 통계치를 이용하여 신호 세기들을 정규화하며, 정규화된 신호 세기들을 이용하여 의료 영상에서 관심 부위를 검출한다. 이에 의해, 병변이나 혈관 구조 등에 대한 정량적 분석과 접근이 용이하고 객관적으로 가능해진다.
의료 영상 비식별화 시스템, 방법 및 컴퓨터 판독 가능한 기록매체 의료 영상 비식별화 시스템, 방법 및 컴퓨터 판독 가능한 기록매체가 개시된다. 본 발명의 일 실시예에 따른 의료 영상 비식별화 시스템은, 환자의 안면 영역을 식별할 수 있는 이미지를 포함하는 의료 영상을 수신하는 의료 영상 수신부, 상기 의료 영상에서 상기 안면 영역을 분리하는 안면 영역 분리부, 클라이언트로부터 상기 안면 영역에 포함되는 눈, 코, 입 중 적어도 어느 하나에 대한 자료 제공 요청을 수신하는 요청 수신부 및 요청된 부위를 제외한 나머지 안면 영역에 대하여 비식별화를 수행하는 비식별화부를 포함한다.
의료용 영상처리장치 본 고안은 의료용 영상처리장치에 관한 것으로, 엑스레이튜브에서 조사되고 인체 등에 투시되는 방사선이 영상증배관에서 증배되고, 상기 증배된 영상을 디지털방식으로 촬영하는 카메라로 촬영하여 디지털 영상정보가 생성되며, 상기 디지털 영상정보가 컴퓨터에 연결된 저장장치에 저장되거나 출력장치를 통해 직접 출력되어 사용되도록 됨은 물론, 상기 영상정보가 국제의료영상표준에 의해 표준화되어 다른 기기 및 사용자와 공유됨과 더불어, 인터넷 등의 통신망을 통해 외부 기기 및 사용자와 공유되도록 된 것이다.
의료 영상 장치 및 의료 영상 처리 방법 혈관을 나타내는 3차원 의료 영상 및 2차원 의료 영상을 획득하는 단계; 2차원 의료 영상 내의 혈관의 일부 영역과 대응되는 3차원 의료 영상 내의 혈관 영역을 결정하는 단계; 및 혈관 영역을 2차원 의료 영상 내의 혈관의 일부 영역에 정합하는 단계를 포함하는 의료 영상 처리 방법 및 이를 위한 의료 영상 장치가 개시된다.
의료 영상 장치 및 의료 영상을 처리하는 방법 의료 영상 장치는 디스플레이, 하나 이상의 인스트럭션을 저장하는 메모리 및 메모리에 저장된 하나 이상의 인스트럭션을 실행하는 프로세서를 포함할 수 있다. 일 실시예에 따른 프로세서는, 하나 이상의 인스트럭션을 실행함으로써, 의료 영상 촬영 데이터에 기초하여 관심 영역(ROI)에 대한 렌더링된 3차원 의료 영상을 생성하고, 3차원 의료 영상에 포함된 적어도 하나의 관심 구조를 인식하고, 적어도 하나의 관심 구조를 나타내는 제1 영상을 포함하는 적어도 하나의 관심 구조 아이콘을 생성하고, 적어도 하나의 관심 구조 아이콘을 디스플레이부에 디스플레이하도록 제어할 수 있다.
의료영상 장치, 영상 처리 장치 및 영상 융합 방법 서로 다른 종류의 영상을 정합함에 있어서, 환자의 특정 움직임을 효과적이고 정밀하게 보정할 수 있는 의료영상 장치, 영상 처리 장치 및 영상 융합 방법을 제공한다. 일 측면에 따른 서로 다른 종류(modality)의 제1영상과 제2영상을 융합하는 의료 영상 장치는, 환자의 주기적인 움직임에 의해 나타나는 객체의 변형에 관한 통계적 객체 모델을 대상 환자의 객체 특성에 맞게 변형한 환자 적응형 객체 모델 및 상기 대상 환자의 제2영상을 저장하는 저장부; 상기 대상 환자의 제1영상을 획득하는 영상 획득부; 및 상기 환자 적응형 객체 모델로부터 추출된 객체의 형상과 상기 제1영상에 나타난 객체의 형상 사이의 오차에 기초하여 상기 제2영상을 변형시키는 영상 융합부;를 포함한다.
의료 영상 장치 및 의료 영상 처리 방법 엑스선 소스로부터 대상체로 조사된 엑스선을 검출하여 획득되는 측정 데이터를 획득하는 데이터 획득부 및 상기 측정 데이터에 기초하여 초기 영상을 획득하고, 상기 측정 데이터 및 상기 초기 영상에 기초하여 ROI(region of interest) 외부 측정 데이터 및 ROI 내부 측정 데이터를 교차로 추정하고, 상기 ROI 내부 측정 데이터에 기초하여 재구성 영상을 획득하는 영상 처리부를 포함하는 의료 영상 장치가 개시된다.
의료용 영상장치 및 그 영상보정 방법 본 발명은 의료용 영상장치 및 그 영상보정 방법을 개시한다. 본 발명의 영상장치는 카메라를 통해 입력되는 영상에 대해 일률적인 자동 노출보정, 자동 초점 보정 및 자동 화이트 밸런스를 수행하는 것이 아니라, 사용자의 관심 영역에 대한 정보를 사용자로부터 획득하여 해당 영역이 최적으로 보상되도록 자동 보정한 영상을 표시할 수 있다. 이에 따라 일률적인 자동 보정에 의해 사용자의 관심 영역의 영상이 식별 불가능하게 되거나 적정 수준으로 보정되지 않는 문제를 해결한다.
2차원 의료영상으로부터 3차원 영상을 생성하는 방법 2D 의료영상이 단 1장만 있는 경우에도 이로부터 간단하게 3차원 의료 영상을 생성할 수 있는 방법. 본 발명의 3차원 영상 생성 방법은 2차원 의료영상을 필터링한 후 각 장기를 나타내는 영역들 각가에 대하여 서로 다른 맵핑 프로파일에 의해 각 화소에 그레이레벨에 따른 깊이 값을 부여함으로써 깊이 맵을 생성하고, 이를 토대로 3차원 영상을 구현한다. 본 발명에 의한 3차원 영상 생성 방법의 한 형태에 따르면, 먼저 소정의 필터링 기법에 의해 상기 2차원 의료영상을 필터링한다. 그리고, 전체 그레이레벨 범위를 구분한 소구간들 각각에 대해 서로 다른 프로파일에 의해 그레이레벨에 깊이 값을 맵핑하는 규칙에 따라서, 필터링된 영상의 각 화소에 상기 화소의 그레이레벨에 따른 깊이 값을 부여함으로써 깊이 맵을 생성한다. 그 다음, 상기 깊이 맵을 2차원 의료영상에 적용하여 스테레오스코픽 영상쌍을 구성하는 좌영상 및 우영상을 생성하고, 좌영상 및 우영상을 포맷팅하여 3차원 스테레오스코픽 영상을 표시한다.
기계학습용 의료 영상 데이터의 학습 노이즈 필터링 방법 및 그 시스템 본 발명은 기계학습에 방해가 되는 노이즈를 학습 목적에 따라 선택적으로 필터링할 수 있는 방법 및 그 시스템에 관한 것으로, 본 발명의 일 실시예에 따른 기계학습용 의료 영상 데이터의 학습 노이즈 필터링 시스템은 기계학습 목적에 따른 학습 노이즈의 정의사항을 수신하는 학습 노이즈 정의부; 상기 학습 노이즈를 필터링하기 위한 필터를 생성하는 필터 생성부; 및 상기 생성된 필터를 이용하여 기계학습에 방해가 되는 학습 노이즈를 제거하는 필터링부를 포함할 수 있다.
정제된 인공지능 강화학습 데이터 생성을 통한 의료영상 판독 시스템 및 그 방법 본 발명은 정제된 인공지능 강화학습 데이터 생성을 통한 의료영상 판독 시스템 및 그 방법에 관한 것으로, 의료영상 판독 전문가의 판독문으로부터 의료영상 판독의 강화 학습 데이터를 추출하여 이를 인공지능 학습 데이터로 활용함으로써 인공지능 의료영상 판독의 계산비용 및 복잡도를 낮추고 정확도를 향상시킬 수 있는 정제된 인공지능 강화학습 데이터 생성을 통한 의료영상 판독 시스템 및 그 방법에 관한 것이다.
캡슐 내시경 영상을 이용한 자동 출혈 감지 방법 및 컴퓨터 프로그램 본 발명은 캡슐 내시경 영상을 이용한 자동 출혈 감지 방법 및 컴퓨터 프로그램에 관한 것으로서, 상기 방법은, 캡슐 내시경 기기로부터 수신한 원본 이미지에 대해 미리 정의된 기준에 따른 이미지 처리를 수행하여 변환된 이미지를 획득하는 이미지 프로세싱 단계; 변환된 이미지로부터 얻은 색조(hue, h), 채도(saturation, s), 명도(value, v)값을 이용하여 미리 정의된 알고리즘의 입력값을 논리 연산하고, 논리 연산을 통해 출혈부 식별 이미지를 획득하는 경계값(threshold) 분석 단계; 및 미리 정의된 필터를 이용하여 상기 출혈부 식별 이미지에 포함된 출혈부를 표시(label)하여 최종 이미지를 획득하는 형태학적 분류 단계를 포함한다.
의료영상 관리 시스템, 방법 및 컴퓨터 판독 가능한 기록매체 의료영상 관리 시스템, 방법 및 컴퓨터 판독 가능한 기록매체가 개시된다. 본 발명의 일 실시예에 따른 의료영상 관리 시스템은, 의료영상의 특성 및 의료영상에 포함된 잡음(artifacts)의 특성에 관한 정보를 포함하는 데이터베이스, 의료영상을 수신하는 영상 수신부, 상기 데이터베이스에 저장된 상기 의료영상의 특성에 관한 정보를 이용하여 수신된 상기 의료영상의 종류를 판단하는 영상 분류부 및 상기 데이터베이스에 저장된 상기 잡음의 특성에 관한 정보를 이용하여 수신된 상기 의료영상의 품질을 판단하는 품질 판단부를 포함한다.
기계학습을 통한 의료영상 판독 및 진단 통합 시스템 본 발명은 기계학습을 통한 의료영상 판독 및 진단 통합 시스템에 관한 것으로, 특정 신체부위에 대한 의료영상에 대해서 기계학습을 수행하여 학습모델을 생성한 다음, 생성한 학습모델을 이용하여 특정 신체부위를 촬영한 의료영상을 쿼리로 입력하여 해당 신체부위에 대한 정상/비정상 판별, 병변위치 추정, 판별된 병변과 비슷한 케이스를 검색하여 제공하고, 신체연령을 추정하여 제시하며, 질병의 시계열적인 변화를 통해서 질병의 현재상태를 진단하고 질병의 미래상태를 예측하며, 현재상태에서 가장 적합한 치료방법을 제시하는 통합 시스템을 제공하는 것에 관한 것이다.
3차원 의료 영상에서 깊이 방향 재귀 학습을 이용하는 병변 특징 표현 분석 방법 및 시스템 일실시예에 따르면, 3차원 의료 영상에서 깊이 방향 재귀 학습을 이용하는 병변 특징 표현 분석 방법은 컨벌루션 신경망(Convolution Neural Network; CNN)을 이용하여, 복수의 단면영상들-상기 복수의 단면영상들은 병변을 촬영한 하나의 3차원 의료 영상에 포함됨- 각각의 병변 공간 특징 표현을 추출하는 단계; 및 상기 복수의 단면영상들 각각의 병변 공간 특징 표현을 기초로 재귀 신경망(Recurrent Neural Network)을 이용하여, 상기 복수의 단면영상들의 깊이 방향에 따라 상기 복수의 단면영상들 각각의 병변 특징 표현을 추출하는 단계를 포함한다.
X-Ray 의료 영상 판독을 위한 데이터 심화학습 처리 기술 및 그 방법 본 발명은 유무선 네트워크를 통해 이루어지고 있는 사물인터넷, 클라우드 등을 통해 다양한 형태의 데이터가 기하급수적으로 증가하고 있으며 공장 자동화, 의료장비 등 다양한 산업분야에서 나타나고 있다. 서로 연결된 정보에서 효과적으로 활용하기 위해 인공지능 기능을 통해 영상 인식, 데이터 분석 등에 활용이 가능하다. 인공지능은 여러 비선형 변환기법의 조합을 통해 추상화를 통한 기계학습 기술을 통하여 심층 신경망회로와 병렬처리 프로세서를 결합한 컴퓨터구조 및 그를 이용한 이미지 변환 및 분석 처리방법에 관한 것으로서, 클라우드 서버를 통해 수집된 다양한 정보, 이미지 등의 정형화된 데이터의 특징점, 색, 선, 형상 등의 정보를 추출하여 상기 물체의 이미지에 대한 2차원 좌표값으로 변환하고, 상기 좌표 값을 패킷의 형태로 출력하는 신경망 회로와, 복수의 프로세싱 유닛을 포함하고, 상기 추출된 특징점을 중심으로 하여 형성된 이미지 패치 내부의 픽셀들 간의 다중 인지 회로 연산을 수행하는 소프트웨어 와 다중 레이어 기반의 데이터를 정형화, 필터철, 비선형화, 추출 등의 과정을 반복적으로 수행함으로써 데이터 및 이미지의 영상 데이터를 분석 처리할 수 있도록 한다.
딥러닝을 이용한 의료영상의 공간 정규화 장치 및 그 방법 의료영상의 공간 정규화 장치는 복수 개의 기능적 의료영상을 심층 학습 아키텍처에 입력하면 미리 저장된 학습데이터에 기초하여 기능적 의료영상을 공간적 정규화하기 위한 적응형 템플릿을 생성하는 적응형 템플릿 생성부, 입력받은 사용자의 기능적 의료영상에 대해서 생성적 적대 신경망(Generative adversarial network: GAN)을 통해 적응형 템플릿에 기초한 이미지를 생성하고 생성된 이미지의 진위 여부를 판단하는 과정을 반복하여 학습하는 학습부, 그리고 학습 결과에 기초하여 공간 정규화된 사용자의 기능적 의료영상을 제공하는 공간 정규화부를 포함한다.
의료 영상 생성 방법 및 디바이스 본 발명은, 조영제가 투여되지 않은 목적 부위에 대한, 비-조영제 의료 영상을 수신하는 단계, 비-조영제 의료 영상 및 목적 부위에 조영제가 투여되어 촬영된 조영제 의료 영상을 기초로 학습되어, 조영제가 투여될 경우 활성화되는 영역을 비-조영제 영상에서 예측하도록 구성된 활성 영역 예측 모델을 이용하여, 비-조영제 의료 영상 내에서 활성 영역을 예측하는 단계, 및 활성 영역이 표시된 비-조영제 의료 영상을 제공하는 단계를 포함하는, 의료 영상 생성 방법 및 디바이스를 제공한다.
의료 이미지 해석을 위한 시스템 및 방법 인공 지능 소견 시스템은, 의료 이미지 데이터를 수신하고, 사용자에 대한 저장된 선택사항을 고려하는 이미지 해석 알고리즘 및 의료 이미지 데이터에 기초하여 소견을 생성하는 소견 엔진을 포함한다. 클라이언트 뷰어는 소견을 사용자에게 제시한다. 조정 엔진은 사용자가 소견을 조정하여 리포트를 생성할 수 있게 한다. 추적 모듈은 리포트를 생성할 때 사용자에 의해 소견에 이루어진 조정 및 소견을 추적한다. 추적 모듈은 추적 정보를 생성한다. 기계 학습 엔진은 추적 정보를 수신하고, 사용자에 의해 소견에 이루어진 조정에 기초하여 사용자에 대한 저장된 선택사항을 조정한다.
개인 맞춤형 뇌질병 진단 및 상태 판정을 위한 의료 영상 처리 시스템 및 방법 본 발명은 비침습적 자기공명영상을 이용하여 영상을 획득하고 뇌조직과 뇌혈관의 복합적 영상 분석을 하여 현재 뇌의 상태와 질병별 위험도, 질병의 예측이 가능하도록 한 개인 맞춤형 뇌질병 진단 및 상태 판정을 위한 의료 영상 처리 시스템 및 방법에 관한 것으로, 3차원 T1 구조 영상(3D T1 Weighted Image), 2차원 T2 FLAIR(2D T2 fluid attenuated inversion recovery) 영상, 뇌혈관 이상을 확인하기 위해 혈관만을 영상화하는 MRA(Magnetic Resonance Angiogram) 영상 및 혈관에 흐르는 혈류의 상태를 파악하기 위한 4차원 위상 대조도 흐름(Phase-contrast Flow) 영상을 획득하는 영상 처리부;진단 대상 질병을 선택하고 그에 따른 뇌 영역을 설정하여 뇌조직(brain tissue), 뇌혈관(vessel) 분석을 수행하는 복합 영상 분석부;연령대별 한국인 데이터 DB를 활용하여 기계학습(machine learning) 알고리즘으로 뇌상태, 질병별 위험도, 발병 가능성, 질병 예측 결과를 출력하는 개인별 진단 및 결과 출력부;를 포함하는 것이다.
의료 검사를 위한 이미지 분석 관리 방법 및 서버 미세 세포를 검사하는 의료 검사에 있어서, 원격의 샘플을 하나의 통합된 검사 장소에서 자동 및 수동으로 검사하고, 이에 대한 검사 결과에 따른 학습이 가능하도록 하여, 자동 의료 검사의 정확도를 높이는 한편, 수기 분석 및 자동 분석의 인력, 시간 및 장소의 제한을 최대한으로 해소할 수 있는 기술을 제공한다. 본 발명의 일 실시예에 따른 의료 검사를 위한 이미지 분석 관리 방법은, 의료 검사를 위한 이미지 분석 관리 서버가, 검사 장소에 마련된 단말로부터 기설정된 의료 검사를 위해 채취된 샘플을 현미경으로 촬영한 이미지를 수신하는 단계; 자동 분석 기기 또는 수기 분석을 수행하는 관리자의 단말로부터, 수신한 이미지를 자동 분석 기기를 통한 분석 및 수기 분석 중 어느 한 분석 방법을 이용하여 분석한 검사 결과 데이터를 수신하는 단계; 및 수신한 검사 결과 데이터를 검사 장소에 마련된 단말에 전송하는 단계;를 포함하는 것을 특징으로 한다.
오브젝트 추적 장치 및 그의 오브젝트 추적 방법 오브젝트 추적 장치의 오브젝트 추적 방법이 개시된다. 본 오브젝트 추적 방법은 영상을 구성하는 복수의 프레임들 중 추적 대상에 대한 추적이 완료된 프레임 집합과 추적이 완료되지 않은 프레임 집합 간에 확률 전파를 수행하여 추적이 완료되지 않은 프레임 집합에 포함된 프레임 각각에서 추적 대상에 대한 확률맵을 산출하는 단계, 산출된 확률맵에 기초하여 추적이 완료되지 않은 프레임 집합에서 하나의 프레임을 선택하는 단계 및, 선택된 프레임에서 추적 대상의 위치를 결정하는 단계를 포함한다.
객체 인식 장치 및 방법 일 실시예에 따른 객체 인식 장치는, 원영상과 상기 원영상의 샘플 분할 맵(segmentation map)을 포함하는 복수의 영상 학습 집합(image learning set)으로부터 특징 추출을 하기 위한 컨볼루션 신경망(convolution network)을 이용하여 상기 원영상의 이미지 특징 벡터를 결정하고, 상기 결정된 이미지 특징 벡터로부터 디컨볼루션 신경망(deconvolution network)을 이용하여 상기 원영상의 분할 맵을 결정하고, 상기 샘플 분할 맵 및 상기 결정된 분할 맵을 이용하여 상기 컨볼루션 신경망의 가중치 및 상기 디컨볼루션 신경망의 가중치를 결정하는 학습부 및 상기 결정된 컨볼루션 신경망의 가중치를 이용하는 컨볼루션 신경망 및 상기 결정된 디컨볼루션 신경망의 가중치를 이용하는 디컨볼루션의 신경망을 통해 입력 영상으로부터 상기 입력 영상의 분할 맵을 결정하는 분할 맵 결정부를 포함할 수 있다.
목표 대상 추적 장치 및 방법 영상의 목표 대상을 추적하는 추적 장치가 목표 대상을 추적하기 위한 방법에 있어서, 목표 대상이 위치한 추적 영역을 포함하는 프레임들에 대한 추적 처리 순서를 나타내는 제 1 트리 구조를 획득하는 단계, 프레임들로부터 두 개의 프레임들로 구성된 복수 개의 프레임 그룹들을 획득하고, 프레임 그룹들 각각에 대하여 거리 평가값들을 획득하는 단계, 제 1 트리 구조와 거리 평가값들에 따라 제 2 트리 구조를 획득하는 단계, 및 획득된 제 2 트리 구조에 따라 목표 대상을 추적하는 단계를 포함하고, 거리 평가값은 프레임 그룹을 구성하는 두 개의 프레임들이 포함된 추적 영역들의 위치 및 추적 영역들에 포함된 픽셀값들 중 적어도 하나에 의해 결정되는 것을 특징으로 하는 추적 방법이 제공된다.
사전 정보를 이용한 영상 물체 탐색 방법 및 이를 수행하는 장치 사전 정보를 이용한 영상 물체 탐색 방법 및 이를 수행하는 장치가 개시된다. 영상 물체 탐색 방법은 구조화 예측 프레임워크에 사전 정보를 결합한 학습 프레임워크를 생성하는 단계, 학습 프레임워크에 대한 교대 최적화 학습을 실행하는 단계, 교대 최적화 학습이 실행된 학습 프레임워크로부터 예측 모델을 생성하는 단계, 및 예측 모델을 이용하여 테스트 샘플이나 입력 영상으로부터의 특정 이미지에서 물체를 예측 혹은 탐색하는 단계를 포함한다.
촬상 장치 및 제어 방법 촬상 장치가 개시된다. 촬상 장치는 외부 영상을 촬상하는 촬상부, 촬상된 이전 영상 프레임 및 현재 영상 프레임을 복수의 영역으로 구분하여 이전 영상 프레임의 각 영역에 대응되는 현재 영상 프레임의 각 영역의 이동 방향 및 거리를 산출하고, 산출된 이동 방향 및 거리를 기초로 기 설정된 배경 모델 및 객체 모델을 적용하여 배경 및 객체를 결정하는 영역 결정부 및 바디 파트 모델을 기초로 결정된 객체의 바디 파트를 판단하고, 판단된 바디 파트를 조합하여 객체의 자세를 결정하는 자세 추정부를 포함한다. 이에 따라, 촬상 장치는 다양한 환경에서 영상 프레임의 배경과 객체를 구분하고 객체의 자세를 판단할 수 있다.
온라인 비디오 분할 방법 및 이를 수행하는 장치 온라인 비디오 분할 방법 및 이를 수행하는 장치가 개시된다. 온라인 비디오 분할 방법은, 제공된 비디오에 포함된 n번째(여기서, n은 1이상의 자연수) 이미지에 속한 화소들을 적어도 하나의 군집으로 군집화하는 단계 및 적어도 하나의 군집 각각에 대해 추출한 색상 히스토그램에 기초하여 각 군집에 대한 병합을 수행하여 n번째 이미지를 복수의 영역으로 분할하는 단계를 포함한다. 따라서, 비디오 상에서 지속적으로 변화하는 군집의 개수를 자동적으로 추정할 수 있고, 이를 통해 비디오 분할의 정확도를 향상시킬 수 있다.
레이어 블러 모델에 기반한 비디오 디블러링 방법, 이를 수행하기 위한 기록 매체 및 장치 레이어 블러 모델에 기반한 비디오 디블러링 방법은, 블러 비디오 프레임을 수신하면, 카메라의 노출시간 동안의 레이어의 결합으로 이루어진 이미지들을 이용하여, 각 프레임의 레이어별 잠상(latent image), 물체의 움직임 및 마스크를 추정하는 단계; 추정된 각 프레임의 레이어별 잠상, 물체의 움직임 및 마스크를 레이어 블러 모델에 적용하여 블러 프레임을 생성하는 단계; 생성된 블러 프레임과 수신된 블러 비디오 프레임을 비교하는 단계; 및 상기 생성된 블러 프레임과 상기 수신된 블러 비디오 프레임이 일치하는 경우, 상기 추정된 각 프레임의 레이어별 물체의 움직임 및 마스크를 기초로, 최종 잠상을 출력하는 단계를 포함한다. 이에 따라, 블러된 영상을 노출 도중의 전경과 배경의 결합으로 이루어진 이미지들의 중첩으로 모델링함으로써, 경계에서의 가리어짐이 표현 가능하여 물체의 경계에서 보다 정확한 디블러링 결과를 얻어낼 수 있다.
3차원 복원을 위한 다중시점 영상 디블러링 방법, 이를 수행하기 위한 기록 매체 및 장치 3차원 복원을 위한 다중시점 영상 디블러링 방법은, 다중시점에서 동기화된 복수의 카메라들에 의해 촬영된 영상들을 수신하는 단계; 상기 수신한 영상들에 대해, 각 시점에서의 깊이 지도(depth map), 레이턴트 영상(latent image) 및 3차원 움직임의 추정을 반복하는 단계; 상기 각 시점에서의 영상 디블러링이 완료되었는지 확인하는 단계; 및 상기 각 시점에서의 최종 깊이 지도(depth map) 및 레이턴트 영상(latent image)를 기초로, 3차원 복원을 수행하는 단계를 포함한다. 이에 따라, 임의의 움직이는 블러 영상에 대해서도 정확한 디블러링 및 3차원 복원의 수행이 가능하다.
영상 상의 물체 추적 방법 본 발명은 영상 상에서 물체의 외양과 움직임에 따라 물체를 추적하는 방법에 관한 것이다. 영상 상의 물체 추적 방법은 영상 상에서 추적하고자 하는 물체에 대한 복수의 기본 외양 모델과 복수의 기본 움직임 모델을 조합하여 복수의 기본 추적기를 형성하는 단계, 그리고 상기 복수의 기본 추적기를 상호 작용시켜 상기 물체를 추적하는 단계를 포함한다. 본 발명에 의하면, 영상 상에서 물체 추적이 어려운 상황들이 동시에 복합적으로 일어날 때에도 정확한 물체 추적이 가능해진다.
카메라 포즈 결정 방법 본 발명은 Special Euclidean Group SE(3) 상에서 PSO(Particle Swarm Optimization)을 수행하여 카메라의 포즈를 결정하는 방법에 관한 것으로, 본 발명에 따른 방법은 이전 프레임에서 카메라의 포즈를 나타내는 복수 개의 파티클(particle)과 상기 복수 개의 파티클 각각을 위한 복수 개의 랜드마크를 초기화하는 단계, 현재 프레임에서 상기 복수 개의 랜드마크에 각각 대응하는 복수 개의 코너 포인트를 추출하는 단계, 그리고 상기 추출된 코너 포인트를 이용하여 상기 복수 개의 파티클에 대해 Special Euclidean Group SE(3) 상에서 PSO(Particle Swarm Optimization)을 수행하여 현재 프레임에 대한 카메라의 포즈를 결정하는 단계를 포함하고, 상기 파티클은 동차 행렬(Homogeneous matrix) 으로 표현되며, R은 3×3 회전 행렬이고, t()는 이동 벡터(translation vector)이다. 본 발명에 따르면 카메라의 급격한 모션 변화에도 카메라 포즈를 종래보다 상대적으로 적은 에러율로 정확하게 구할 수 있다.
영상의 모션 블러를 제거하는 방법 및 장치 본 발명은 카메라에서 획득한 영상의 모션 블러(motion blur)를 제거하는 영상의 모션 블러를 제거하는 방법으로, 이 방법은 카메라의 움직임으로부터 영상의 모션 블러 양상을 예측하는 단계, 모션 블러 양상을 이용하여 이전 영상과 영상 간의 대응 관계 정보를 검출하는 단계, 대응 관계 정보를 이용하여 카메라의 위치를 검출하는 단계, 그리고 카메라의 위치를 이용하여 모션 블러 함수를 추정하고, 모션 블러 함수를 이용하여 영상의 모션 블러를 제거하는 단계를 포함한다.
객체/배경 분리 및 영상분할을 이용한 양안/다시점스테레오 정합 장치 및 그 방법 본 발명은 객체/배경 분리 및 영상분할을 이용한 양안/다시점 스테레오 정합 장치 및 그 방법에 관한 것으로, 영상 장치로부터 획득된 양안/다시점 영상을 객체 및 배경 영역으로 구분하고, 각 구분된 영역별로 양안/다시점 영상을 영상 분할하여 분할된 영역(Segment)을 기반으로 스테레오 정합함으로써, 차폐 영역에 의한 오정합을 줄일 수 있으며 그로 인해 더욱 정확한 변이지도를 얻을 수 있게 하는, 객체/배경 분리 및 영상분할을 이용한 양안/다시점 스테레오 정합 장치 및 그 방법을 제공하고자 한다. 이를 위하여, 본 발명은 양안/다시점 영상의 영역을 객체 영역 및 배경 영역으로 각각 구분하여 판단하고, 상기 판단 결과에 따라 상기 양안/다시점 영상에 대한 객체 및 배경 분리 정보를 생성하기 위한 영역 판단 수단; 상기 양안/다시점 영상을 유사 영역별로 영상분할하고, 상기 영상분할된 양안/다시점 영상과 상기 생성된 객체 및 배경 분리 정보를 이용하여 객체 변이지도 및 배경 변이지도를 개별적으로 구하기 위한 변이 추정 수단; 및 상기 객체 변이지도와 상기 배경 변이지도를 합성하여 상기 양안/다시점 영상에 대한 장면 변이지도를 생성하기 위한 변이지도 합성 수단을 포함한다.
실시간 확장 가능한 스테레오 매칭 시스템 및 방법 본 발명은 실시간 확장 가능한 스테레오 매칭 시스템으로서, 상세하게는 두 개의 카메라를 통해서 얻은 영상을 이용하여 유사한 점을 찾아 그 결과로 깊이(Depth) 영상을 계산하는 스테레오 매칭(Stereo Matching)을 실시간으로 수행하는 실시간 확장 가능한 스테레오 매칭 시스템에 관한 것이다. 본 발명의 실시예에 따른 실시간 확장 가능한 스테레오 매칭 시스템은 도 3에 나타낸 바와 같이, 좌표 생성 모듈부(110), 센서스 변환 모듈부(120), 지연 XOR 연산 모듈부(130), 스테레오 매칭 모듈부(140), 제어 모듈부(150)로 구성된다. 본 발명에 따른 실시간 스테레오 매칭 시스템을 이용하면, 별도의 컴퓨터 시스템이나 소프트웨어 프로그램을 사용하지 않고 실시간으로 교정된 영상의 깊이 정보를 획득하는 것이 가능하며 또한, 발명된 장치는 하드웨어로 간단하게 구현될 수 있어 로봇 비전(Robot Vision)이나 지능형 카메라(Intelligent Camera), 산업현장 등에 쉽게 응용될 수 있다.
실시간 스테레오 영상 교정 시스템 및 방법 본 발명은 실시간 스테레오 영상 교정 시스템 및 방법에 관한 것으로, 상세하게는 두 개의 카메라에 측정(Camera Calibration)을 수행하여 얻은 측정 행렬(Calibration Matrix)을 사용하여 카메라에서 얻은 영상을 실시간으로 교정(Rectification)하는 실시간 스테레오 영상 교정 시스템 및 방법에 관한 것이다. 본 발명은 좌표 생성 모듈부(110), 교정 좌표 생성 모듈부(120), 이중 선형 보간 값 생성 모듈부(130), 교정 좌표 메모리(140), 이중 선형 보간 메모리(150), 영상 버퍼(160), 교정 모듈부(170), 제어 모듈부(180)로 구성된다. 본 발명에서는 실시간으로 스테레오 교정을 수행할 수 있는 하드웨어 시스템의 구조를 제공하였으며, 하드웨어 장치를 통해 동작과 연산 결과를 검증하였다. 그리고, 실시간 스테레오 교정 시스템을 이용하면, 별도의 컴퓨터 시스템이나 소프트웨어 프로그램을 사용하지 않고 실시간으로 교정된 영상을 획득하는 것이 가능하다.
지능형 영상 감시 시스템에서의 물체 추적 방법 및 장치 본 발명은 (a) 입력 영상을 입력받는 단계와, (b) 상기 입력 영상에서 이동 물체의 선형적인 움직임을 고려하여 상기 입력 영상의 전체 영역에 걸쳐서 상기 이동 물체를 추정하는 단계와, (c) 상기 이동 물체의 비선형적인 움직임을 고려하여 추정된 상기 이동 물체의 주변 영역에 걸쳐서 상기 이동 물체를 추정하는 단계를 구비하는 지능형 영상 감시 시스템의 물체 추적 방법 및 장치를 제공한다.
점진적 학습 기반의 물체 추적 시스템에서의 데이터 선택방법 및 장치 본 발명은 점진적 학습 기반의 물체 추적 시스템에서의 데이터 선택 방법 및 장치에 관한 것으로, 본 발명의 일 실시 예에 따른 방법은 외양 모델을 갱신하기 위한 추적 데이터들과 비교 추적 데이터들을 비교하고, 비교 결과에 따라 상기 추적 데이터들 중에서 일부를 선택함으로써, 외양 모델 갱신에 보다 적합한 추적 데이터를 선택하여 추적 성능을 향상시킨다.
히스토그램 기반 및 픽셀 기반의 물체 추적 방법 및 장치 본 발명의 물체 추적 방법은, 템플릿의 각 픽셀들에 관한 히스토그램이 전경일 확률 값을 기초로 템플릿 가중 이미지를 추정하는 단계, 후보 이미지의 각 픽셀들에 관한 히스토그램이 전경일 확률 값을 기초로 후보 가중 이미지를 추정하는 단계, 템플릿 가중 이미지 및 후보 가중 이미지를 기초로, 후보 이미지가 템플릿과 유사할수록 0에 가까운 값을 갖는 가중 비율을 연산하는 단계, 가중 비율 및 템플릿의 픽셀과 후보 이미지의 픽셀의 차이에 기초한 차제곱합(SSD) 항을 가지는 목적 함수를 최적화하는 움직임 벡터를 연산하는 단계 및 목적 함수를 최적화하는 움직임 벡터에 상응하는 후보 이미지로써 템플릿을 갱신하는 단계를 포함할 수 있다.
듀얼 이벤트 카메라를 이용한 SLAM 방법 및 SLAM 시스템 SLAM(simultaneous localization and mapping) 방법은 듀얼 이벤트 기반 카메라를 채용한다. 카메라로부터의 이벤트 스트림은 환경 내의 표면 포인트를 입체적으로(stereoscopically) 검출하고, 카메라가 움직임에 따라 카메라의 포즈를 동적으로 계산하고, 동시에, 환경의 맵을 업데이트한다. 기울기 하강(gradient descent) 기반 최적화가 각각의 이벤트 또는 각각의 작은 이벤트 배치(batch)에 대하여 포즈를 업데이트하는데 활용될 수 있다.
자동 로컬리제이션을 위한 장치 및 방법 적어도 하나의 메모리 요소를 포함하는 이동 로봇이 개시된다. 적어도 하나의 메모리 요소는, 적어도 하나의 지도 데이터; 시각적 이미지들을 촬영하도록 구성된 적어도 2대의 카메라; 및 적어도 로봇을 로컬리제이션하기 위해, 적어도 2대의 카메라에 의해 촬영된 시각적 이미지들로부터 적어도 직선들을 추출하여 추출된 직선들을 지도 데이터와 비교하도록 구성된 적어도 하나의 처리 요소를 포함한다. 적어도 2대의 카메라로 시각적 이미지들을 촬영하는 단계; 적어도 하나의 처리 요소를 갖는 개별적인 시각적 이미지로부터 직선들을 추출하는 단계; 추출된 특징들을 기존의 지도 데이터와 비교하는 단계; 및 전술한 비교에 기초하여 위치 가설을 출력하는 단계를 포함하는 로컬리제이션 방법이 더 개시된다.
복수 개의 깊이 영상에 기초한 깊이 영상 접힘 제거 장치 및 방법 깊이 영상에서 깊이 접힘을 제거하여 3차원 구조를 복원하는 장치 및 방법을 제공한다. 깊이 영상 접힘 제거 장치는 고정된 깊이 영상 촬영장치의 변조 주파수를 달리하여 촬영한, 동일한 장면(scene)에 대한 복수 개의 깊이 영상들을 입력 받는 입력부, 상기 복수 개의 깊이 영상들 각각에서 상기 장면의 동일한 위치를 나타내는 픽셀의 3차원 점간 거리에 기초하여 깊이 접힘 횟수를 추정하는 깊이 접힘 횟수 추정부 및 상기 추정된 깊이 접힘 횟수에 기초하여 상기 복수 개의 깊이 영상에서 깊이 값의 접힘이 제거된 복수 개의 깊이 영상을 출력하는 출력부를 포함한다.
모션 블러 제거 방법 및 이러한 방법을 수행하는 장치 모션 블러 제거 방법 및 이러한 방법을 수행하는 장치가 개시되어 있다. 비균일 모션 블러를 제거하는 방법은 비균일 모션 블러 영상을 기초로 제1 레이턴트 영상을 예측하는 단계, 비균일 모션 블러 영상 및 상기 제1 레이턴트 영상을 정합하여 호모그라피 및 상기 호모그라피에 대한 가중치를 포함하는 비균일 모션 블러 정보를 추정하는 단계와 추정된 비균일 모션 블러 정보를 기초로 비균일 블러가 제거된 제2 레이턴트 영상을 추정하는 단계를 포함할 수 있다. 따라서, 비균일 모션 블러가 포함된 영상에서 비균일 모션 블러를 추정하여 제거할 수 있고, 종래의 비균일 모션 블러를 제거할 수 없는 블러 예측 및 제거 방법보다 정확하게 영상에 발생한 비균일 모션 블러를 추정하여 블러가 발생한 영상을 더 선명하게 복원할 수 있다.
이미지 생성 방법 및 시스템 본 발명은 이미지 생성 방법 및 시스템에 관한 것으로, 특히 적어도 두 개의 이미지를 결합하여 출력하기 위한 이미지 생성 방법 및 시스템에 관한 것이다. 이를 달성하기 위해 이미지 생성 방법이, 적어도 하나의 정의된 영역을 가지는 제1 입력이미지를 포함하는 탬플릿을 생성하는 단계; 제2 입력 이미지를 수신하는 단계; 상기 적어도 하나의 정의된 영역에 포함된 복수의 픽셀들과 관련된 세기를 상기 제2 입력 이미지에 포함되는 복수의 픽셀들에 적용하여 상기 제2 입력이미지를 변경하고, 상기 적어도 하나의 정의된 영역을 상기 변경된 제2 입력이미지로 교체하여 출력이미지를 생성하는 단계 및 상기 출력이미지를 디스플레이하는 단계를 포함하는 것을 특징으로 한다.
머신 러닝을 위한 학습 이미지 생성 장치 및 방법 본 발명은 머신 러닝(machine learning)용 학습을 위한 무기 체계 이미지를 생성하는 장치 및 방법에 대한 것으로, 특정 무기 체계에 대한 간접 정보, 및 배경 이미지를 입력받는 입력부와, 상기 특정 무기 체계에 대한 간접 정보에 근거하여 상기 특정 무기 체계의 3차원 객체인 무기 체계 객체를 생성하는 모델링부와, 상기 무기 체계 객체를 적어도 하나의 방향에서 바라보는 적어도 하나의 객체 이미지와 상기 배경 이미지를 합성하여, 적어도 하나의 합성 이미지를 생성하는 합성부와, 상기 적어도 하나의 합성 이미지를 저장하는 메모리 및, 상기 특정 무기 체계의 제원에 따라 상기 무기 체계 객체의 크기를 조절하고, 상기 크기가 조절된 무기 체계 객체에 대한 상기 적어도 하나의 객체 이미지를 생성 및, 상기 배경 이미지 상에서 사용자가 선택한 위치에 상기 객체 이미지가 표시되는 상기 적어도 하나의 합성 이미지를 생성하도록 상기 합성부를 제어하며, 상기 저장된 합성 이미지를 상기 특정 무기 체계에 대한 머신 러닝(machine learning) 학습 데이터로 상기 메모리에 저장하는 제어부를 포함하는 것을 특징으로 한다.
얼굴에 대응하는 3차원 아바타를 이용하여 얼굴의 움직임이 반영된 3차원 아바타를 포함하는 이미지를 생성하는 전자 장치 및 그 동작 방법 다양한 실시예에 따른 전자 장치는, 카메라, 디스플레이, 및 프로세서를 포함하고, 상기 프로세서는, 상기 카메라를 이용하여, 하나 이상의 외부 객체들에 대한 제1 이미지를 획득하고, 상기 제1 이미지에 포함된 상기 하나 이상의 외부 객체들 중 얼굴에 대응하는 외부 객체와 관련된 속성에 기반하여 생성된 3 차원 그래픽 객체를 상기 디스플레이를 통해 제공하고, 상기 3 차원 그래픽 객체에 적용될 수 있는 복수의 그래픽 속성들 중 선택된 적어도 하나의 그래픽 속성 및 상기 3차원 그래픽 객체에 적어도 기반하여, 상기 외부 객체에 대한 3 차원 아바타를 생성하고, 및 상기 3 차원 아바타를 이용하여 지정된 표정 또는 모션이 반영된 적어도 하나의 객체를 포함하는 제2 이미지를 생성하도록 설정될 수 있다. 그 밖의 다양한 실시예가 가능하다.
딥스테레오: 실세계 이미지로부터 새로운 뷰들을 예측하는 러닝 기존의 이미지들로부터 새로운 뷰들을 예측하기 위해 딥 네트워크들(deep networks)을 사용하는 딥 러닝(deep learning) 시스템 및 방법은 대규모 데이터로부터 모델들 및 표현들을 생성하고 개선한다. 이 딥러닝 시스템 및 방법은 수 많은 포즈의 이미지 세트들로부터 트레이닝된 픽셀들로부터 직접적으로 새로운 뷰 합성을 수행하는 딥 아키텍처(deep architecture)를 이용한다. 이 타입의 딥 네트워크를 이용하는 시스템은 이웃하는 뷰들의 픽셀들에 기초하여 보이지 않는 뷰의 픽셀들을 생산할 수 있으며, 그 자체를 그래픽 생성시 어플리케이션에 제공한다.
무인항공기 기반의 무인차량 주행 방법 및 무인차량 주행 시스템 무인항공기 기반의 무인차량 주행 방법 및 무인차량 주행 시스템이 개시된다. 무인항공기와 통신하여 자율 주행하는 무인차량 주행 방법에 있어서, 상기 무인항공기가 상기 무인차량의 주행 환경 정보를 획득하고, 상기 주행 환경 정보에 따라 상기 무인차량이 주행하여야 하는 최적 경로 정보를 생성하여 상기 무인차량에 전송하고, 상기 무인차량이 상기 최적 경로 정보를 수신하고, 상기 최적 경로 정보에 따라 자율 주행하는 무인차량 주행 방법을 제공한다.
차량의 자율주행 시스템 및 방법 본 발명은 자율주행차량 시스템 및 방법에 관한 것으로, 차량의 전방, 후방, 좌측 및 우측 차량의 제동 능력 및 무게를 확인하여 차량이 사고가 발생하는 것을 방지할 수 있도록 하는 자율주행 차량 시스템 및 방법에 관한 것이다. 보다 상세하게는 차량 종류에 따른 제동 능력 정보를 제정하고, 운행중인 차량의 전방 및 후방에서 주행중인 차량의 종류를 확인하고, 차량 정보 데이터베이스를 참조하여 주변 차량 정보 확인부에서 확인한 전방 및 후방에서 주행중인 차량의 종류에 대응되는 제동 능력 정보를 확인하고, 확인된 제동 능력 정보를 기초로 현재 차선의 사고 발생 위험 등급을 계산하고, 사고 위험 계산부에서 계산된 사고 발생 위험 등급을 기초로 차선 변경 여부를 판단할 수 있도록 하는 기술이다.
자율 주행 차량, 자율 주행 관리 장치 및 그 제어 방법 본 발명은 자율 주행 차량, 자율 주행 관리 장치 및 그 제어 방법에 관한 것으로서, 본 발명의 일 실시예에 따른 자율 주행 차량은, 상기 차량의 위치 정보를 수신하는 무선 통신부, 전자지도를 저장하는 저장부, 목적지를 입력받는 입력부, 주행 속도 및 주행 방향을 조절하는 구동부 및 상기 전자지도에서 상기 위치 정보에 대응하는 지점부터 상기 목적지에 대응하는 지점까지의 경로를 탐색하고, 상기 무선 통신부를 통해 상기 경로에 포함된 적어도 하나 이상의 구간에 대한 점유권을 승인받으면, 상기 구동부를 통해 상기 차량이 상기 승인된 구간을 주행하도록 제어하는 제어부를 포함한다.
자율주행 차량과 사용자 간의 인터랙션 시스템 및 방법 본 발명에 따른 복수의 주행모드를 지원하는 자율주행 차량과 운전자 간의 인터랙션 방법은 운전자로부터 입력된 목적지 정보 및 상기 자율주행 차량의 현재위치 정보에 따른 상기 주행모드 별 예상 주행경로 및 예상 주행시간 정보를 획득하는 단계 및 상기 복수의 주행모드 중 운전자로부터 선택된 주행모드에 대한 예상 주행경로 및 예상 주행시간 정보를 운전자에게 안내하는 단계를 포함한다.
무인 자율 이동체 및 무인 자율 이동체의 적외선 통신 방법 무인 자율 이동체는 무인 자율 이동체를 이동시키는 구동부, 무인 자율 이동체의 현재 위치를 측정하는 위치 측정부, 위치 측정부에서 측정된 현재 위치를 기초로 구동부를 제어하여 무인 자율 이동체의 주행을 제어하는 주행 제어부, 제동 신호와 연동되며 제동 신호가 수신되는 경우 무인 자율 이동체의 후방으로 적외선 신호를 송신하는 적외선 송신부 및 구동부, 위치 측정부, 주행 제어부 및 적외선 송신부를 제어하여 무인 자율 이동체의 주행을 제어하는 중앙 처리부를 포함한다.
차량간 통신을 이용한 자율주행 자동차의 차선 변경 장치 및 방법 본 발명의 목적은, 자율주행 자동차들 간에 통신을 수행하여 차선 변경을 수행할 수 있는, 차량간 통신을 이용한 자율주행 자동차의 차선 변경 장치 및 방법을 제공하는 것이다. 이를 위해, 본 발명에 따른 차량간 통신을 이용한 자율주행 자동차의 차선 변경 장치는, 자율주행 자동차의 운행 정보를 감지하는 감지부; 상기 운행 정보를 무선 통신망을 통해 주기적으로 출력하며 또 다른 자율주행 자동차의 운행 정보를 수신하는 통신부; 및 차선 변경이 필요할 때, 상기 운행 정보들을 이용하여 차선 변경 요청 메시지를 상기 통신부를 통해 출력하며, 차선 변경 승인 메시지의 수신 여부에 따라 차선 변경 여부를 결정하는 제어부를 포함한다.
자율주행 전기자동차의 제어방법 본 발명은 자율주행 전기자동차의 제어방법에 관한 것으로, 차체에 구비된 신호 수신부가 도로에 설치된 복수의 신호 발신부 중 어느 하나의 상측에 위치하는 경우 그 신호 발신부와 일대일 무선통신하여 신호상태, 시간정보, 거리정보, 그리고 차선정보를 수신하는 교차로 정보 수신단계와, 내비게이션으로부터 목적지 도착을 위한 주행방향, 주행도로의 규정속도, 그리고 현재 주행속도를 제공받는 운행상태 파악단계와, 상기 신호 수신부와 상기 내비게이션으로부터 수신한 정보와 현재 운행상태를 비교하여 주행방향과, 주행동력을 발생시키는 구동모터를 동작시켜 주행속도를 제어하는 주행제어단계, 그리고 주행제어에 대응하여 상기 구동모터에 전원을 공급하는 배터리부를 냉매의 잠열을 이용하여 냉각시키는 냉각수단을 제어하는 배터리부 냉각단계;를 포함한다. 이러한 방법으로, 교차로에서 안전하게 차량의 주행을 제어할 수 있고, 배터리부를 효과적으로 냉각하여 배터리부의 성능 유지 및 수명을 연장할 수 있는 효과를 얻을 수 있다.
자율주행 전기자동차 본 발명은 자율주행 전기자동차에 관한 것으로, 상기 자율주행 전기자동차는 차체에 구비되고, 도로에 설치된 복수의 신호 발신부 중 어느 하나의 상측에 위치하는 경우 그 신호 발신부와 일대일 무선통신하여 신호상태, 시간정보, 거리정보, 그리고 차선정보를 수신하는 신호 수신부와, 목적지 도착을 위한 주행방향, 주행도로의 규정속도, 그리고 현재 주행속도를 제공하는 내비게이션과, 배터리부로부터 전원을 공급받아 주행동력을 발생시키는 구동모터와, 상기 신호 수신부와 상기 내비게이션으로부터 수신한 정보와 현재 운행상태를 비교하여 주행방향과 상기 구동모터를 동작시켜 주행속도를 제어하는 제어부, 그리고 상기 배터리부를 냉매의 잠열을 이용하여 냉각하는 냉각수단을 포함한다. 이러한 구성으로, 교차로에서 안전하게 차량의 주행을 제어할 수 있고, 배터리부를 효과적으로 냉각하여 배터리부의 성능 유지 및 수명을 연장할 수 있는 효과를 얻을 수 있다.
전기자동차용 자율주행 제어시스템 본 발명은 전기자동차용 자율주행 제어시스템에 관한 것으로, 거리정보와 차선정보가 저장된 신호 발신부와, 신호상태와 시간정보를 상기 신호 발신부에 전송하는 신호등과, 통행정보를 상기 신호 발신부에 전송하는 교차로 감지부와, 보행정보를 상기 신호 발신부에 전송하는 횡단보도 감지부와, 신호 발신부와 일대일 무선통신하여 정보를 수신하는 신호 수신부와, 주행방향과 규정속도 그리고 현재 주행속도를 제공하는 내비게이션과, 배터리부로부터 전원을 공급받아 상기 전기자동차의 주행동력을 발생시키는 구동모터와, 상기 전기자동차의 주행방향과 상기 구동모터를 동작시켜 상기 전기자동차의 주행속도를 제어하는 제어부, 그리고 상기 배터리부를 냉매의 잠열을 이용하여 냉각하는 냉각수단;을 포함한다. 이러한 구성으로, 교차로에서 안전하게 차량의 주행을 제어할 수 있고, 배터리부를 효과적으로 냉각하여 배터리부의 성능 유지 및 수명을 연장할 수 있는 효과를 얻을 수 있다.
전기자동차용 자율주행 제어시스템의 제어방법 본 발명은 전기자동차용 자율주행 제어시스템의 제어방법에 관한 것으로, 거리정보와 차선정보가 저장된 신호 발신부가 신호등으로부터 신호상태와 시간정보를 수신하여 저장하는 신호등 정보 수신단계와, 상기 신호 발신부가 교차로 감지부로부터 통행정보를 수신하여 저장하는 통행정보 수신단계와, 상기 신호 발신부가 횡단보도 감지부로부터 보행정보를 수신하여 저장하는 보행정보 수신단계와, 신호 수신부가 신호 발신부와 일대일 무선통신하여 정보를 수신하는 교차로 정보 수신단계와, 내비게이션으로부터 주행방향과 주행도로의 규정속도 그리고 현재 주행속도를 제공받는 운행상태 파악단계와, 상기 전기자동차의 주행방향과 주행속도를 제어하는 주행제어단계, 그리고 배터리부를 냉매의 잠열을 이용하여 냉각시키는 배터리부 냉각단계를 포함한다. 이러한 방법으로, 교차로에서 안전하게 차량의 주행을 제어할 수 있고, 배터리부를 효과적으로 냉각하여 배터리부의 성능 유지 및 수명을 연장할 수 있는 효과를 얻을 수 있다.
자율주행 차량의 시뮬레이션을 위한 테스트정보 제공 방법 및 장치 자율주행 차량의 시뮬레이션을 위한 테스트정보 제공 방법 및 장치가 제공된다. 본 발명의 일 실시예에 따른 자율주행 차량의 시뮬레이션을 위한 테스트정보 제공 장치는 차량에 탑재되는 각종 센서로부터 획득된 일정 시간의 센싱정보로부터 고정 물체와 이동 물체를 포함하는 주변 물체들을 추출하는 주변 물체 추출부; 상기 추출된 주변 물체들에 대하여 하나 이상의 체크포인트(check point)가 설정되면, 상기 체크포인트에 대한 설정 정보를 상기 센싱정보와 결합하여 테스트정보를 생성하는 테스트정보 생성부; 및 상기 테스트정보를 자율주행을 위한 제어명령을 출력하는 제어모듈로 제공하는 테스트정보 제공부;를 포함하는 것을 특징으로 한다.
자율주행 차량 운용 관리 제어 자율주행 차량 운용 관리는 자율주행 차량에 의해 차량 운송 네트워크를 횡단하는 것을 포함할 수 있다. 차량 운송 네트워크를 횡단하는 것은 자율주행 차량의 센서로부터, 자율주행 차량의 정의된 거리 내에 있는 외부 물체에 대응하는 센서 정보를 수신하는 것, 센서 정보를 수신하는 것에 응답하여 별개의 차량 운용 시나리오를 식별하는 것, 시나리오 특정 운용 제어 평가 모듈 인스턴스를 인스턴스화하는 것 - 시나리오 특정 운용 제어 평가 모듈 인스턴스는 별개의 차량 운용 시나리오를 모델링하는 시나리오 특정 운용 제어 평가 모듈의 인스턴스임 -, 시나리오 특정 운용 제어 평가 모듈 인스턴스로부터 후보 차량 제어 액션을 수신하는 것, 및 후보 차량 제어 액션에 기초하여 차량 운송 네트워크의 부분을 횡단하는 것을 포함할 수 있다.
주행 경험 정보를 이용한 자율주행 학습 장치 및 방법 본 발명은 주행 경험 정보를 이용한 자율주행 학습 장치에 관한 것으로, 주행중인 차량의 주행 환경 정보를 수집하는 환경정보 수집 센싱부; 주행중인 차량의 거동 제어 정보를 수집하는 제어정보 수집 센싱부; 및 수집되는 거동 제어 정보를 기준으로 자차 주변의 변화하는 주행 환경 정보를 매칭하여 주행 경험 정보를 생성하는 자율주행정보 생성부;를 포함한다.
자율주행 차량의 오탑승 방지 방법 및 그 장치 자율주행 차량의 오탑승 방지 방법 및 그 장치가 개시된다. 본 발명의 일 실시예에 따른 자율주행 차량의 오탑승 방지 방법은 탑승자 단말로 호출한 제1 차량의 위치 정보와 탑승자의 단말의 위치 정보를 바탕으로 탑승자 단말의 위치가 지속적으로 변동되며, 제1 차량의 위치와 탑승자 단말의 위치가 미리 정해진 거리 이상으로 멀어지는 경우, 탑승자가 호출하지 않은 제2 차량에 탑승한 것으로 판단하고 환승 여부를 문의하는 메시지를 탑승자 단말을 통해 표시하는 것이다. 본 발명에 의하면 다른 차량을 탑승하고 간 탑승자를 기다리지 않아도 되어 효율적이며, 차량 안의 범죄를 빨리 추적할 수 있다. 본 발명의 자율 주행 차량이 인공 지능(Artificial Intelligence) 모듈, 드론(Unmanned Aerial Vehicle, UAV), 로봇, 증강 현실(Augmented Reality, AR) 장치, 가상 현실(virtual reality, VR) 장치, 5G 서비스와 관련된 장치 등과 연계될 수 있다.
자율주행 차량의 운행 모드 전환을 위한 장치, 이를 위한 방법 및 이 방법을 수행하는 프로그램이 기록된 컴퓨터 판독 가능한 기록매체 본 발명은 자율주행 차량의 운행 모드 전환을 위한 장치, 이를 위한 방법 및 이 방법을 수행하는 프로그램이 기록된 컴퓨터 판독 가능한 기록매체에 관한 것으로, 이러한 본 발명은 상기 차량의 운전자를 촬영하는 카메라부와, 상기 차량의 외부를 복수의 영역으로 구분하고, 구분된 복수의 영역의 객체를 감지하는 복수의 센서를 포함하는 센서부와, 상기 차량의 종방향 및 횡방향 운행을 제어하기 위한 복수의 계통을 제어하는 구동부와, GPS(Global Positioning System) 신호, DGPS(Differential GPS) 신호 및 관성센서 신호를 이용하여 상기 차량의 위치 정보를 도출하는 위치정보부와, 상기 카메라부 상기 센서부, 상기 구동부, 및 상기 위치정보부를 포함하는 상기 차량을 제어하는 2 이상의 프로세서를 포함하는 제어부를 포함하며, 상기 제어부는 상기 차량의 주행이 상기 프로세서에 의해 제어되는 완전자율주행 레벨 및 상기 차량의 주행이 운전자의 조작 및 상기 프로세서 양자에 의해 제어되는 반자율주행 레벨 중 어느 하나의 자율주행 레벨에 따라 자율주행을 수행하는 자율주행 모드에서 상기 센서부, 상기 구동부, 상기 위치정보부 및 상기 제어부 중 적어도 하나의 기능이 정상 상태가 아닌 패일 상태인지 여부를 판단하고, 상기 패일 상태이면, 재부팅을 수행하는 차량장치와, 이를 위한 방법 및 이 방법을 수행하는 프로그램이 기록된 컴퓨터 판독 가능한 기록매체를 제공한다.
차량의 자율주행 강화학습 장치 및 방법 본 발명은 차량의 자율주행 강화학습 장치 및 방법에 관한 것으로, 상기 장치는, 차량의 자율주행 중 운전자의 조작 개입을 인식하는 운전자 개입 인식부; 상기 차량의 상태 정보 및 차량 주변의 상황 정보를 복수의 센서를 이용해 검출하는 차량정보 검출부; 교통 인프라 혹은 다른 차량과 통신하여 차량의 주행과 정지 및 차선 변경을 위한 교통정보를 수신하고, 강화학습을 통해 자율주행 알고리즘 개선 동작을 수행하는 서버와 통신하여 이 개선 동작에 필요한 정보를 전송하며, 개선 완료된 자율주행 알고리즘을 상기 서버로부터 전달받는 통신부; 차량의 자율주행 중 운전자의 조작 개입 시, 상기 운전자 개입 인식부를 통해 인식한 운전자의 개입 조작 정보와 자율주행 알고리즘에 의해 판단한 제어 정보를 비교하여 상기 자율주행 알고리즘의 개선이 필요한지 판단하고, 개선이 필요할 경우에 상기 통신부를 통해 개선 동작에 필요한 정보를 전송하게 하는 제어부; 및 상기 통신부를 통해 유무선 방식으로 전달되는 정보를 이용해 강화학습을 통해 자율주행 알고리즘 개선 동작을 수행하는 상기 서버;를 포함한다.
자율주행 차량의 배터리 충전 방법 및 이를 위한 장치 본 명세서는 자율주행 시스템에서 충전 차량에 의해서 수행되는 배터리 충전 방법을 제공한다. 보다 구체적으로, 충전 차량은 관제 센터로부터 주행 차량의 배터리 충전을 요청하는 요청 메시지를 수신하되, 상기 요청 메시지는 상기 주행 차량의 위치와 관련된 제 1 위치 정보 및 상기 주행 차량의 배터리 잔량과 관련된 배터리 정보를 포함한다. 이후, 충전 차량은 상기 관제 센터로부터 상기 주행 차량의 배터리를 충전하기 위한 충전 정보를 포함하는 제어 메시지를 수신하되, 상기 충전 정보는 상기 배터리 충전을 위한 도로 위치와 관련된 도로 위치 정보, 충전을 위해 점유하는 점유 차선과 관련된 차선 정보, 및 상기 점유 차선의 점유 시간과 관련된 시간 정보를 포함한다. 이후, 충전 차량은 상기 제 1 위치 정보에 기초하여 상기 주행 차량과 인접한 위치로 이동하고, 상기 충전 정보에 기초하여 상기 점유 차선에서 상기 주행 차량의 배터리를 충전한다. 본 발명의 자율 주행 차량, 사용자 단말기 및 서버 중 하나 이상이 인공 지능(Artificail Intelligenfce) 모듈, 드론(Unmanned Aerial Vehicle, UAV), 로봇, 증강 현실(Augmented Reality, AR) 장치, 가상 현실(virtual reality, VR) 장치, 5G 서비스와 관련된 장치 등과 연계될 수 있다.
자율주행차량의 센서 고장 처리 장치, 시스템 및 방법 본 발명은 자율주행차량의 센서 고장 처리 장치, 시스템 및 방법에 관한 것으로, 자율주행차량이 자율주행을 실시하기 위해 필요한 적어도 하나 이상의 정보를 검출하는 적어도 하나 이상의 센서를 포함하는 센서부; 상기 센서부의 적어도 하나 이상의 센서에 대한 고장 여부를 체크하고, 주변차량이나 서버로부터 고장난 센서에 대한 대체센서정보를 수신하여 자율주행을 위한 연산을 수행하는 제어부; 및 상기 제어부의 제어에 따라, 상기 주변차량이나 서버로부터 대체센서정보를 요청하거나 수신하는 통신부;를 포함한다.
자율 주행 차량 시스템 및 차량의 자율 주행 방법 본 발명은 서버로부터 수신되는 서비스 데이터에 기초하여 제1 신호를 생성하는 통신 장치; 생성되는 센싱 데이터에 기초하여 제2 신호를 생성하는 복수의 센서; 생성되는 차량의 위치 데이터에 기초하여 제3 신호를 생성하는 위치 측위 장치; 상기 제1 신호, 상기 제2 신호 및 상기 제3 신호 중 적어도 어느 하나에 기초하여, 주행 상황을 인지하고, 상기 주행 상황에 기초하여, 자율 주행 경로를 생성하고, 상기 자율 주행 경로에 따른 주행을 위한 제어 파라미터에 대한 제4 신호를 생성하는 자율 주행 장치; 및 상기 제4 신호가 수신되지 않는 경우, 상기 제1 신호에 따라, 적어도 하나의 차량 구동 장치를 제어하는 제어 장치;를 포함하는 자율 주행 차량 시스템에 관한 것이다.
자율주행 지원 장치 및 방법 본 발명은 자율주행을 위한 도로정보의 생성, 오차 보정 및 도로 정보의 업데이트 수행할 수 있는 자율주행 지원 장치 및 방법에 관한 것이다. 본 발명의 실시 예에 따른 자율주행 지원 장치는, 자율주행 차량에 배치된 센서부 및 서버를 포함한다. 센서부는 주행 도로를 센싱하여 도로 정보, 표지판 정보, 신호등 정보, 공사구간 정보, 미래 경로 정보를 포함하는 카메라 인식 정보를 생성한다. 서버는 적어도 하나의 자율주행 차량의 센서부에서 수신된 적어도 하나의 카메라 인식 정보를 분석하여 기 설정된 도로맵 정보를 업데이트하고, 업데이트된 도로맵 정보를 상기 적어도 하나의 자율주행 차량으로 전송한다.
자율주행 차량을 위한 기계학습 데이터 표준화 방법 및 시스템 본 발명은 자율주행 차량을 위한 기계학습 데이터 표준화 방법 및 시스템에 관한 것으로서, 더욱 상세하게는 자율 주행 차량이 주변의 상황을 인식하여 자율주행을 수행함에 있어서 기계학습의 데이터를 격자화 및 표준화 하여 연산부하를 감소시키면서 학습을 수행할 수 있는 방법 및 시스템에 관한 것으로, 자율 주행 차량의 주변 환경을 센싱해 인지정보를 획득하는 인지정보 제공부, 상기 인지정보를 코드화 하는 인지정보 통합 모듈, 및 상기 코드화된 인지정보를 이용해 격자형맵을 생성하는 격자형 맵 생성 모듈을 포함하는 구성을 개시한다.
자율주행 차량의 판단 지능 향상을 위한 주행상황 데이터 공유 및 학습 장치 및 그 동작 방법 본 발명은 자율주행 차량의 판단 지능 향상을 위한 주행상황 데이터 공유 및 학습 장치 및 그 동작 방법에 관한 것으로서, 본 발명의 실시예에 따른 자율주행 차량의 판단 지능 향상을 위한 주행상황 데이터 공유 및 학습 장치는 자율주행 차량을 기준으로 기설정된 소정 거리 이내에서 주행중인 주변 차량을 감지하는 센싱부, 상기 자율주행 차량과 타차량 간, 또는 상기 자율주행 차량과 클라우드 서버 간에 데이터를 송수신하는 통신부, 차선 레벨(Lane-lever)의 정밀 맵 데이터를 저장하는 저장부, 및 상기 센싱부의 감지 결과에 대한 주행상황 데이터를 상기 정밀 맵 데이터에 매핑하여 상기 자율주행 차량 중심의 매핑 데이터를 생성하며, 상기 매핑 데이터를 상기 통신부를 통해 상기 타차량 또는 상기 클라우드 서버로 전송하며, 상기 매핑 데이터 및 상기 타차량 또는 상기 클라우드 서버로부터 수신되는 데이터를 이용하여 자율주행을 위한 학습을 수행하는 학습부를 포함한다.
자율주행 자동차를 조종하는 방법 본 발명은 2 개의 운전 모드 중 어느 하나에 따라 자동차(10)를 운전하는 수단(13, 14, 15)을 조종하기 위한 방법에 관한 것으로, 운전 모드는: 운전하는 수단이 자동차 운전자(20)에 의해 수동으로 제어되는 수동 운전 모드 및 운전하는 수단이 자동차의 연산 유닛(12)에 의해 자동으로 제어되는 자율주행 운전 모드이다. 본 발명에 따르면, 자율주행 모드의 종료시: 자동차 전방의 주변 영역의 적어도 하나의 이미지를 획득하는 단계, 획득된 이미지 내의 주변 영역의 적어도 일부의 가시성 계수를 계산하는 단계, 및 가시성 계수에 따라, 적어도 2 개의 별개의 종료 방식에서 자율주행 운전 모드를 종료하는 방식을 선택하는 단계가 제공된다.
자율주행 차량의 주행정보 제공 방법 자율주행 차량에 구비된 센서를 통해 외부상황 데이터를 획득하는 단계, 상기 외부상황 데이터에 기초하여 상기 자율주행 차량, 상기 자율주행 차량으로부터 임계거리 이내에 있는 외부차량 및 상기 자율주행 차량이 주행 중인 도로를 포함하는 로컬 맵을 생성하는 단계 및 상기 로컬 맵을 상기 자율주행 차량의 표시부에 디스플레이하는 단계를 포함하는 자율주행 차량의 주행정보 제공방법이 개시된다.
자율주행 가능구간 예측 시스템 및 그 동작 방법 본 발명은 자율주행 가능구간 예측 시스템 및 그 동작 방법에 관한 것으로서, 본 발명의 실시예에 따른 자율주행 가능구간 예측 시스템은 다수의 자율주행차량 중 임의의 자율주행차량이 진입하고자 하는 도로구간에 대한 자율주행 가능 여부를 판단하고, 상기 판단 결과인 자율주행 가능 여부 판단 정보를 상기 임의의 자율주행차량으로 전송하는 서버, 및 상기 임의의 자율주행차량에 구비되며, 상기 임의의 자율주행차량의 주행모드에 대한 정보를 상기 서버로 전송하고, 상기 서버로부터 수신되는 상기 자율주행 가능 여부 판단 정보를 확인하고, 상기 임의의 자율주행차량의 주행모드와 상기 진입하고자 하는 도로구간에 대한 자율주행 가능 여부를 고려하여 상기 임의의 자율주행차량의 동작을 선택적으로 제어하는 자율주행장치를 포함한다.
자율주행 시스템에서 도커 제어 방법 및 그 장치 자율주행 시스템에서 도커 제어 방법 및 장치가 개시된다. 본 발명의 일 실시예에 따른 것을 특징으로 하는 자율주행 시스템에서 도커 제어 방법 은, 도커(docker)를 요청하는 차량 호출 메시지를 탑승객의 디바이스로부터 수신하는 경우, 탑승객의 호출 위치를 기반으로 차량을 선정하고, 상기 차량 호출 메시지에 대한 응답으로 도커 요청 메시지를 도커 플랫폼으로 전송한다. 또한 도커 플랫폼으로부터 이용 가능한 도커 리스트를 수신한 경우, 도커 리스트 내에 포함된 도커들 중에서 특정 도커를 선택하여 도킹 서비스의 제공을 요청하는 도커 호출 메시지를 선택된 도커로 전송할 수 있다. 이에 따라 탑승객의 수화물 발생 지점과 가장 가까운 도커를 호출함으로써 사용자 편의성을 향상시킬 수 있다. 본 발명의 자율주행 시스템에서 제공하는 제어 방법은, 인공 지능(Artificail Intelligenfce) 모듈, 드론(Unmanned Aerial Vehicle, UAV), 로봇, 증강 현실(Augmented Reality, AR) 장치, 가상 현실(virtual reality, VR) 장치, 5G 서비스와 관련된 장치 등과 연계될 수 있다.
인공지능 자율주행차의 운행시스템 중앙처리장치 본 발명은 복수의 인공지능 자율주행차가 동시에 동일한 선택을 했을때 발생할 수 있는 차량사고의 방지를 위한 상호간의 무선통신 및 주행방법의 통제장치 시스템에 관한 것이다. 더욱 구체적으로는, 주행중인 복수의 인공지능 자율주행차가 동일한 상황에서 동일한 판단을 하게되면 사고가 발생하게 된다. 예를 들어 주행간에 장애물이 나타났을 때 인공지능 자율주행차는 각자가 최선의 판단을 하게 되고 만약 그 각자의 판단이 같다면 같은 경로로 회피하여 운행하게 되고 결과적으로 충돌사고가 발생하게 된다. 즉 인공지능 자율주행차 각자의 최선의 판단이 동일 경로를 주행중인 복수의 인공지능 자율주행 자동차 전체의 최선의 판단은 아니라는 뜻이다. 상기의 문제점을 방지하기 위해 주변에서 주행중인 복수의 자율주행차 상호간에 무선통신을 하는 장치를 두고 인공지능 자율주행 시스템의 판단을 상호 교류하여 다른 복수의 자율주행차와 다른 경로로 주행할수 있게 만드는 인공지능 자율주행차 운행 통제장치 시스템에 관한 것이다.
의료진단장치의 동작방법 및 의료진단장치 본 발명은 사용자로부터 입력되는 측정 개시명령에 응답하여 기설정된 진단시스템 환경에 기초하여 측정을 실시하되, 상기 진단시스템 환경에 대한 변경 명령이 입력되는 경우에만 이를 반영하여 변경된 진단시스템 환경에 따라 측정을 실시하는 것을 특징으로 하는 의료진단장치의 동작방법 및 의료진단장치에 관한 것이다. 본 발명에 따른 의료진단장치의 동작방법 및 의료진단장치는 측정이라는 의료진단장치 본연의 목적에 충실하도록 작업 흐름을 배치하여, 측정 또는 진단에 소요되는 시간 및 사용자의 조작오류의 빈도를 감소시키고 사용자에게 작업의 편의를 제공할 수 있는 특징이 있다.
원격 의료 진단 시스템 및 방법 본 발명은 샘플을 수용하여 생물학적, 화학적 또는 생화학적 반응을 수행하고, 바코드 또는 RF IC를 구비하는 바이오 디스크 또는 바이오 칩; 상기 바이오 디스크 또는 바이오 칩에서 수행된 반응을 분석하고, 상기 바코드 또는 RF IC를 읽는 리더기를 구비하여 서버의 연결 유무에 관계없이 상기 바이오 디스크 또는 바이오 칩을 인증하거나 측정 데이터를 RF IC에 기록하는 바이오 분석 장치; 상기 사용자 단말기상에 소프트웨어 형태로 상주하여 상기 바이오 분석 장치의 사용 방법을 안내하거나 지시하고 사용자와의 상담 서비스를 제공하는 가상 의사; 의료 전문가 또는 가상 의사와의 상담 서비스를 제공하는 사용자 단말기; 사용자와의 상담 서비스를 제공하는 의료 전문가 단말기; 및 상담 전문가로서 정기 검진 기간 동안에는 의료 전문가를 사용자에게 연결하고 그 외 기간 동안에는 가상 의사를 사용자에게 연결하고 사용자가 의료 전문가와 상담해야 하는 정기 검진 기간을 경과한 경우 사용자와 가상 의사와의 상담 서비스 연결을 차단하는 원격 진단 서버를 포함하는 원격 의료 진단 시스템, 및 이를 이용한 원격 의료 진단 방법을 제공한다.
의료진단 시스템 본 발명은 환자는 의료치료기기를 통해 병원을 직접 내방하지 않고 자신이 원하는 장소 및 시간에 건강상태를 측정할 수 있고, 의료기관 서버에서 검출된 최적정도에 따라 의료치료기기가 재설정됨으로써 검진절차가 획기적으로 간편해짐과 동시에 치료효율을 높일 수 있는 의료진단 시스템에 관한 것이다.
자동위치조정이 가능한 X선디텍터를 갖는 의료용진단기기 본 발명은 X선 디텍터가 상하로 구동됨은 물론 상하구동과 동시에 전후진 구동되어 X선 디텍터를 자동으로 위치조정을 할 수 있도록 구성하여 동물이나 환자가 눕는 진단용 테이블과 함께 사용하더라도 한 번의 동작으로 X선 디텍터의 위치를 조정하여 사용이 편리하게 구성된 의료용 진단기기에 관한 것으로서, 복수의 승강레일이 마련된 X선 촬영스탠드와, 상기 X선 촬영스탠드의 승강레일을 따라 상하로 구동되는 연결부와, X선 수상면을 가지는 X선 디텍터가 장착되며 상기 X선 디덱터를 전후로 구동시키는 슬라이드부와, 상기 슬라이드부에 고정되며 상기 연결부에 연결되어 상기 슬라이드부를 회전구동시키는 회전구동부를 구비하며, 상기 연결부와 상기 회전구동부는 상기 연결부가 상기 X선 촬영스탠드의 저면으로부터 일정위치를 지날 때에 상기 연결부의 승강구동과 상기 회전구동부의 회전구동이 동시에 구동될 수 있도록 설정되어, 한 번의 동작으로도 진단용 테이블의 하부측에 X선 디텍터를 자동으로 위치조정을 할 수 있고, X선 디텍터를 전후로 슬라이드 동작이 가능하도록 구성하여 X선 조사부의 회전동작에 의한 촬영 등이 보다 용이하고 간편하다.
다중 모달리티 영상 생성 기법을 이용한 의료영상진단시스템 본 발명은 다중 모달리티 영상 생성 기법을 이용한 의료영상진단시스템에 관한 것으로, 실제 영상획득장치에서 얻어진 의료영상으로 같은 부위에 대한 다중 모달리티(Modality) 영상을 생성하여 대상 질병 진단에 사용하기 위한 것이다. 이를 위하여 본 발명은 의료영상장비에서 획득된 2차원 또는 3차원 실제 영상데이터를 입력받아 같은 부위에 대한 다른 모달리티 영상을 2차원 또는 3차원으로 생성하는 다중 모달리티 영상 생성용 인공신경망, 의료영상장비에서 획득된 2차원 또는 3차원 실제 영상데이터와 다중 모달리티 영상 생성용 인공신경망에서 생성된 다중 모달리티 영상데이터를 입력받아 다중 모달리티 영상들을 이용해 실제 영상데이터의 질병 여부를 판단하여 출력하는 다중 모달리티 기반 진단용 인공신경망을 포함하여, 실제 영상획득장치에서 얻어진 의료영상을 인공신경망에 입력하여 같은 부위에 대한 다른 모달리티(Modality)의 영상을 생성하고 생성된 모달리티 영상을 입력 영상과 결합하여 다중 모달리티 영상을 구성함으로서 대상 질병을 정확하게 진단할 수 있게 한다.
의료진단장치의 동작방법 및 의료진단장치 본 발명의 실시예는 의료진단장치에 관한 것이다. 본 발명의 실시예에 따른 의료진단장치는 대상체로부터 반사되어 온 수신신호의 주파수 대역을 나누고 상기 주파수 대역을 나눈 상기 수신신호에 대하여 각각의 대역에 맞는 위상 회전을 시키는 빔 포머를 포함하는 것을 특징으로 한다.
임상 정보를 반영한 의료 진단 방법 및 이를 이용하는 장치 임상정보를 반영하고 기계학습 알고리즘을 활용하는 의료영상 진단 방법 및 장치가 개시된다. 진단 방법은, 의료영상을 이용하는 진단 방법으로서, 2차원 영상 또는 3차원 영상을 획득하는 단계, 2차원 영상 또는 3차원 영상을 변환한 2차원 영상을 전처리하는 단계, 전처리된 의료영상에서 제1 특징정보를 추출하는 단계, 획득한 임상정보를 전처리하는 단계, 전처리된 임상정보에서 제2 특징정보를 추출하는 단계, 제1 특징정보와 제2 특징정보를 결합 및 적재하는 단계, 및 적재된 특징 정보들을 기계학습 알고리즘에 입력하여 진단하는 단계를 포함한다.
의료 진단시스템 및 그 동작제어방법 본 발명은 휴대용 통신모듈과의 무선통신을 담당하고, 상기 휴대용 통신모듈과의 무선통신 상태에 관한 정보를 제공하는 무선통신부와; 상기 무선통신부로부터의 무선통신 상태에 관한 정보에 응답하여 시스템을 제어하는 제어부를 포함하되, 상기 제어부는 상기 무선통신 상태에 관한 정보에 응답하여, 시스템의 정상 동작 모드 중 무선 통신 단절 상태가 기설정된 제 1 시구간 동안 계속되면 시스템을 대기모드에 진입하도록 제어하는 의료 진단시스템에 관한 것이다.
통합의료진단 시스템 및 방법 본 발명은 통합의료진단 시스템 및 방법에 관한 것으로서, 의뢰시스템이 인터넷을 통하여 판독시스템에 판독을 의뢰하기 위한 자궁경부 사진영상데이터 및 태아의 초음파 동영상데이터를 서버시스템으로 전송하고, 판독시스템이 서버시스템에 접속하여 영상데이터를 수신한 후, 영상데이터를 판독하여 서버시스템으로 전송하고, 의뢰시스템 및 확인시스템이 서버시스템에 접속하여 영상데이터에 대한 진단결과를 확인하는 통합의료진단 시스템 및 방법에 관한 것이다.
의료진단용 영상 촬영 장치 본 발명은 의료진단용 영상 촬영 장치에 관한 것으로서, 전면에 위치한 렌즈를 이용해 영상을 촬영하고, 후면에 정보 입력 및 정보 표시를 위한 터치스크린이 위치하며, 촬영된 영상을 무선 통신을 통해 전송하는 카메라, 렌즈의 둘레 측으로부터 피사체 방향으로 신장된 원통 형상으로서, 피사체 방향 끝단에는 피사체를 조광하기 위한 조명원이 구비된 경통부, 및 카메라의 하단에 결합된 손잡이 형상으로서, 내부에 조명원으로 전원을 공급하기 위한 배터리를 포함하고, 전원선을 이용해 배터리의 전원을 조명원으로 공급하는 손잡이부를 포함하며, 이를 통해 촬영자가 한 손을 이용해 촬영 장치를 제어하면서 피 진단자를 용이하게 촬영하고 촬영된 영상을 무선 통신을 통해 원격지로 전송할 수 있다.
전후진 및 회전동작이 가능한 테이블을 가지는 의료용진단기기 본 고안은 스탠드형 X선 디텍터 및 이와 함께 사용되는 테이블을 전후진 동작 및 회전가능하게 구성하여 좁은 공간에서도 사용이 편리하게 구성된 전후진 및 회전동작이 가능한 테이블을 가지는 의료용 진단기기에 관한 것이다. 본 고안에 의한 의료용 진단기기는, X선 촬영스탠드와, 환자용 테이블을 포함하여 구성되는 의료용 진단기기에 있어서, 상기 X선 촬영스탠드는 기기본체의 축을 중심으로 회전시키는 회전구동모터와, 회전각도와 위치를 검지하는 포텐셔미터를 더욱 포함하여 구성되며, 상기 환자용 테이블은, 바닥 레일부와, 상기 바닥 레일부에 설치되어 상기 레일부 상을 전후이동하는 하부 회전부와, 상기 하부 회전부 상에 설치되는 회전몸체부와, 상기 회전 몸체부 상에 설치되는 상부 구동부와, 상기 상부 구동부 상에 설치되며 테이블을 고정시키는 테이블 고정판을 더욱 포함하여 구성된다. 본 고안의 기기는, X선 디텍터를 회전가능하게 설치한 스탠드형 기기본체를 회전가능하게 구성하고, 환자를 눕히는 테이블을 레일을 따라 전후진 작동이 가능함과 동시에 회전동작이 가능하도록 구성함으로써 좁은 공간에서도 테이블의 전후진 동작 및 회전 동작이 가능하다.
의료영상 진단 시스템용 영상 전송 장치 및 방법 본 발명은 영상 전송 장치를 개시한다. 보다 상세하게는, 본 발명은 컴퓨팅 장치 및 의료영상 진단장치 등에서 통신단자가 아닌 영상 출력단자를 통해 출력되는 영상을 모니터 이외에 타 표시수단을 통해 표시할 수 있도록 하는 의료영상 진단 시스템용 영상 전송 장치 및 방법에 관한 것이다. 본 발명의 실시예에 따르면, 의료영상 진단장치에 탑재되는 영상 출력 단자 및 USB 단자와 연결되어 각각 영상 신호와 부가 정보를 수신 및 복호화하고, 이를 결합하여 부호화 처리하고 패킷화하여 정보통신망을 통해 송출함으로써, 모바일 단말을 통해 끊김없는 의료진단 영상을 제공하는 효과가 있다.
원격의료진단시스템과원격의료영상전송및수신방법 본 발명은 원격 의료진단시스템과 원격 의료영상 전송 및 수신방법에 관한 것으로서, 전송지에서 X 레이 촬영기, 자기공명장치, 단층촬영기 혹은 초음파 진단기로부터 의료영상을 획득하는 의료영상 획득부, 의료영상 획득부에서 생성된 의료영상을 컴퓨터 내부형식으로 변환하는 영상변환부, 영상변환부에서 변환된 의료영상에 관심영역을 표시하기 위한 관심영역 표시부, 관심영역 표시부에서 관심영역이 표시된 의료영상을 ATM 전송망을 통해 전송하기에 적합한 형태로 변환하고 전송 규약에 맞추어 전송하는 영상전송부, ATM 전송망을 통해 전송된 의료영상을 전송 규약에 맞추어 수신하고, 수신된 전송데이타를 컴퓨터 내부형식으로 변환하는 영상수신부, 영상수신부에서 수신된 영상데이타를 화면에 디스플레이하는 디스플레이부를 구비한다. 따라서, 환자의 의료영상을 전송하는 통신망으로 ATM 전송망을 사용하므로 모든 종류의 영상을 전송시작과 거의 동시에 원격지로 전송이 가능하여 응급상황에 빠르게 대처할 수 있으며, 병원내에서 사용할 경우 쉽게 의료영상 전산화를 이룰 수 있다.
원격진료를 위한 의료영상 처리장치 및 이를 포함한 원격 의료진단 시스템 원격진료를 위한 의료영상 처리장치 및 이를 포함하는 원격 의료진단 시스템이 개시된다. 본 발명의 일 실시 예에 따른 의료영상 처리장치는 의료장비로부터 원격 영상진단장치에 전송되는 의료영상을 대상으로 진단유형 별 영상 특성 및 네트워크 상태에 따라 의료영상의 품질 파라미터를 조정하여 의료영상을 다양한 진단환경에 따라 최적화한다.
인공지능 기반 의료용 자동 진단 보조 방법 및 그 시스템 인공지능 진단을 기반으로 하는 의사결정 지원 시스템이나 컴퓨터 보조 진단 시스템에 전문가 분석을 결합하여 진단 결과의 신뢰성을 높이고 인공지능의 의료용 자동 진단 보조 성능을 향상시킬 수 있는 인공지능 기반 의료용 자동 진단 보조 방법 및 시스템이 개시된다. 호스트에 접속되는 단말 또는 컴퓨팅 장치가 수행하는 인공지능 기반 의료용 자동 진단 보조 방법은, 호스트에서 인공지능 기반으로 진단된 제1 진단 결과를 획득하는 단계, 제1 진단 결과를 포함한 이미지 형태의 사용자 인터페이스를 단말 또는 컴퓨팅 장치의 화면에 표시하는 단계, 사용자 인터페이스를 통해 제1 진단 결과에 대한 전문가 진단을 포함한 사용자 입력을 획득하는 단계, 사용자 입력을 반영한 제2 진단 결과를 생성하는 단계, 및 제1 진단 결과에 대응하는 제2 진단 결과를 호스트에 제공하는 단계를 포함한다.
의료진단 및 원격통신 기능을 갖는 액정표시장치 개시된 본 발명은 소변을 수집하여 분석할 수 있는 장치에 액정표시장치를 연결시켜 소정 사용자가 자신의 소변 분석정보를 직접 확인할 수 있도록 하는 한편, 이상이 발견되는 경우 원거리에 있는 클리닉 센터로 소변 분석정보를 전송하여 전문의의 진단을 받을 수 있도록 하는 의료진단 및 원격통신 기능을 갖는 액정표시장치에 관한 것으로서, 본 발명은 사용자의 소변을 수집하여 사용자의 소변 성분을 분석하는 분석수단; 및 상기 분석수단으로부터 입력되는 소변 성분을 이용하여 사용자의 건강상태를 분석하고, 분석 결과를 도식화하여 화면상에 출력하는 한편, 상기 분석결과를 미리 등록되어 있는 기준 전송 데이터와 비교하여, 상기 기준 전송 데이터를 초과하는 경우 미리 설정 등록된 사용자 단말기의 번호를 이용하여 상기 분석결과를 전송하는 액정표시수단을 포함하여 구성된 것을 특징으로 한다.
의료진단을 위한 의료영상 처리장치 및 그 방법 의료진단을 위한 의료영상 처리장치가 개시된다. 본 발명의 일 실시 예에 따른 의료영상 처리장치는 대상체로부터 반사된 초음파 신호로부터 채널 데이터를 획득하는 프런트 앤드 프로세서와, 프런트 앤드 프로세서를 통해 획득된 채널 데이터를 대상으로 영상 파라미터를 조작하여 초음파 영상을 생성하는 영상 생성부와, 영상 생성부를 통해 생성된 초음파 영상을 분석하여 병변을 검출하는 영상 분석부와, 영상 분석부의 분석 결과를 제공하는 표시부를 포함한다.
진단의료장비용 에어 매트리스 진단의료장비용 에어 매트리스가 개시된다. 본 발명에 의한 진단의료장비용 에어 매트리스는, 검사대상자를 수용하는 튜브 본체와, 튜브 본체에 공기를 주입하거나 배출시키는 공기주입구를 포함하며, 튜브 본체를 방사선 진단의료장비의 테이블 상에 설치하여 검사대상자와 테이블 사이에 공기층을 형성시키는 것을 특징으로 한다. 본 발명에 따르면, 테이블 상에 에어 매트리스를 설치함으로써 공기층(Air gap)에 의한 방사선 피폭선량을 감소시킬 수 있다. 또한, 공기층을 갖는 에어 매트리스의 완충 기능을 통해 등이 굽은 검사대상자, 욕창 환자 등에게 검사시 편안함을 제공할 수 있다.
의료 영상 진단 장치 및 방법 진단 정확도 향상을 위한 의료 영상 진단 장치 및 방법에 관한 기술로, 일 실시 예에 따른 의료 영상 진단 장치는 의료 영상 기기로부터 입력되는 프레임을 샘플링하여 일정 크기의 공간 내 하나 이상의 프레임을 저장하는 저장부, 기준 프레임이 결정되면, 저장된 프레임들 중 기준 프레임의 진단에 이용할 하나 이상의 프레임을 수집하는 프레임 수집부 및 수집한 각 프레임의 진단 결과를 활용하여 기준 프레임을 진단하는 진단부를 포함할 수 있다.
충돌 방지 스위치의 스위치 작동 조절 장치 및 이를 채용한 의료용 진단기기용 테이블 충돌 방지 스위치의 스위치 작동 조절 장치 및 이를 채용한 의료용 진단기기용 테이블이 개시된다. 개시된 의료용 진단기기용 테이블은 지지대와, 지지대에 길이 방향으로 이동 가능하게 지지되는 테이블 프레임과, 테이블 프레임의 양 측단 중 적어도 일 측단에 마련된 가이드 프레임과, 가이드 프레임에 인접하여 지지대에 마련되며 테이블 프레임이 이상 동작시 테이블 프레임의 움직임을 정지시키는 충돌 방지 스위치와, 스위치 작동 조절 장치와, 충돌 방지 스위치의 눌림이 해제되면 상기 테이블 프레임의 움직임을 정지시키는 제어부를 포함하며, 스위치 작동 조절 장치는 가이드 프레임의 변형에 따라 움직이는 제1 레버와, 정상 상태에서 상기 충돌 방지 스위치를 누르고 있으며 제1 레버의 상방의 움직임이 기설정된 높이를 벗어난 경우 제1 레버에 의해 들리어 충돌 방지 스위치의 눌림을 해제하여 충돌 방지 스위치를 작동시키는 제2 레버를 구비한다.
이비인후과용 무선 의료 진단 장치 본 발명은 이비인후과용 무선 의료 진단 장치에 관한 것으로서, 본 발명에 따른 이비인후과용 무선 의료 진단 장치는 이미지센서와 조명광원 및 상기 이미지센서에서 획득한 영상이미지를 영상표시기로 송출하기 위한 무선통신모듈이 마련되고, 일단부에는 제1결합부가 마련된 본체부; 및 신체의 구멍 내부에 위치한 피검부와 상기 이미지센서 사이의 광 경로를 확보하기 위해 신체의 구멍을 통해 삽입되고, 상기 본체부에 조립되는 진단용 삽입기구부;를 포함하는 것을 특징으로 한다.
의료 영상 및 진단을 위한 비디오 클립 선택 장치 클립 선택 방법은 포유류 신체의 표적 부분 및 진단 절차의 선택을 의료 영상 기기에 대한 인터페이스를 통해 검색하는 단계를 포함한다. 이후, 기기는 표적 부분의 비디오 클립 이미지를 획득하고 상기 비디오 클립 이미지를 이미지 저장소에 저장한다. 비디오 클립 이미지의 각 비디오 클립은 이후 각 비디오 클립의 뷰 및 품질을 결정하기 위해 이미지 처리되고, 규칙은 선택된 진단 절차 및 표적 부분에 대응하는 규칙 기반으로부터 검색된다. 이와 관련하여, 규칙은 비디오 클립 이미지의 필수 뷰 및 품질을 지정한다. 마지막으로, 검색된 규칙은 지정된 필수 뷰 및 품질을 만족시키는 비디오 클립 이미지의 부분 집합을 생성하기 위한 필터로서 비디오 클립 이미지에 적용되고, 비디오 클립 이미지의 부분 집합은 이미지 저장소에 저장된다.
의료 영상 이미징 방법, 그에 따른 의료 진단 장치 및 그에 따른 기록 매체 촬영 대상 영역 내에 포함되는 적어도 하나의 화상 프레임을 획득하는 데이터 획득부 및 획득된 적어도 하나의 화상 프레임을 처리하여 의료 영상을 획득하는 데이터 처리부를 포함하는 의료 진단 장치가 개시된다. 의료 진단 장치의 데이터 처리부는, 적어도 하나의 화상 프레임을 이용하여 컴포지트 화상을 생성하는 컴포지트 화상 생성부 및 적어도 하나의 화상 프레임 및 컴포지트 화상에 포함되는 영상 신호에 대응하는 경계 조건을 이용하여 컴포지트 화상을 보상하고, 보상된 컴포지트 화상을 이용하여 의료 영상을 획득하는 의료 영상 획득부를 포함한다.
의료 진단 리포트 생성 장치 및 방법 본 발명은 의료 진단 리포트 생성 장치로, 대상체 병변의 진행 정도 또는 위치를 판단한 진단 결과를 출력하는 컴퓨터 보조 진단(Computer Aided Diagnosis: 이하 'CAD'로 기재함) 시스템이 진단한 동일 의료 영상을 입력받아 진단 결과의 의사 결정 근거가 되는 병변 형상 속성별 평가치를 예측하는 인공 지능 학습 모델 기반의 진단 근거 정보 추출부와, 병변 형상 속성별 평가치에 상응하는 의료 진단 설명문을 생성하는 문장 생성부를 포함한다.
의료 진단 장치 본 발명의 일 측면에 따르면, 피진단자의 상반신을 커버하도록 형성되는 커버부, 및 커버부의 일면에 피진단자의 상반신과 대향하도록 배치되어 피진단자의 생체 음향을 취득하는 복수의 마이크 어레이를 포함하고, 마이크 어레이는 선택적으로 온오프되어 상이한 체구를 갖는 복수의 피진단자로부터 각각 생체 음향을 취득하는 복수의 마이크 유닛을 포함하는 의료 진단 장치가 제공된다.