Abstract:
본 발명은 삼각형 메쉬로 표현된 3차원 물체를 DoSurface 표현 방법으로 변환하는 시스템 및 그 방법에 관한 것으로서, DoSurface 표현 방법의 특성을 이용하여 각 삼각형이 만날 수 있는 직선들의 후보들을 효과적으로 선별하고, 교점을 쉽게 계산할 수 있는 좌표계로 변환을 수행하며, 변환된 좌표계에서 하나의 삼각형과 만날 수 있는 후보 직선과 교점을 빠르게 구할 수 있는 시스템 및 그 방법을 제공함에 목적이 있다. 이러한 목적을 달성하기 위한 본 발명은, 정규평면을 정의하는 정규평면 정의부; DoSurface 표현 방법을 정의하는 직선들을 분류하는 방향벡터 분류부; 분류된 직선들의 방향벡터가 정규평면의 법선 벡터와 일치하도록 변환하는 변환식들을 생성하는 정규평면 변환부; 삼각형 메쉬로 표현된 3차원 물체를 상기 변환식들을 이용하여 좌표계 변환을 수행하는 좌표계 변환부; 변환 좌표계에서 삼각형 내부를 지나는 직선들을 선별하는 후보 직선 선별부; 및 상기 변환 좌표계에서 직선의 시작점에서 삼각형과 직선이 만나는 점까지의 거리를 계산하는 깊이값 계산부; 를 포함한다.
Abstract:
본 발명은 번호판 검출을 위한 영상 압축 시스템 및 그 방법에 관한 것으로서, 차량 영상의 압축을 수행함에 있어, 중요 정보를 포함하는 번호판 영역을 타 영역과 상이한 방식으로 압축함으로써, 차량 영상에서 관심 영역인 번호판 영역의 손실을 최소화할 수 있으며, 복호화 후 번호판 인식의 성능을 향상시킬 수 있도록 함에 그 목적이 있다. 이러한 목적을 달성하기 위한 본 발명은, 번호판 영역이 포함된 차량 영상에서, 번호판 영역을 추출하는 번호판 영역 추출부; 추출된 번호판 영역 정보를 예측하는 번호판 영역정보 예측부; 예측된 번호판 영역 정보와, 예측된 번호판 영역 및 실제 번호판 영역정보와의 잔차(residual) 신호를 압축하는 번호판 영역 압축부; 및 상기 차량 영상의 비 번호판 영역을 압축하는 비 번호판 영역 압축부; 를 포함하되, 상기 번호판 영역 및 비 번호판 영역은, 상이한 방식으로 압축되는 것을 특징으로 한다.
Abstract:
PURPOSE: A 3D data format producing/consuming apparatus and a method thereof can extract information of images, form it in structurized 3D data format, and create a realistic panoramic image. CONSTITUTION: An image preprocessor (110) checks a matching point between images photographed with multiple cameras. The image preprocessor extracts at least one of depth value, texture value, and object segmentation information between images as image information. A video information structurizing unit (120) structurizes 3D data to use image information in producing a realistic image. A 3D data format DB unit (130) stores formatting information of 3D data to in a database. A solid sense video generator (140) uses formatting information of 3D data and produces a realistic panorama. A realistic image renderer (150) renders a realistic panorama. [Reference numerals] (100) 3D data format generation/consumption device; (110) Image preprocessor; (120) Video information structuring unit; (130) 3D data format DB unit; (140) Realistic image video generator; (150) Realistic image renderer
Abstract:
PURPOSE: A system and a method for converting a 3D object represented by triangle meshes into a 3D object represented by depth on surface (DoSurface) are provided to classify straight lines defining DoSurface representation and perform calculation after mapping the straight lines to a normal plane, thereby simplifying a process to acquire intersections between triangles and straight lines. CONSTITUTION: A system for converting a 3D object represented by triangle meshes into a 3D object represented by DoSurface comprises a normal plane definition unit (100) to define a normal plane; a direction vector classification unit (200) to classify straight lines defining DoSurface representation; a normal plane conversion unit (300) to generate a conversion formula in which a direction vector of the classified straight lines is equal to a normal vector of the normal plane; a coordinate system conversion unit (400) to perform conversion of the coordinate system of a 3D object represented by triangle meshes using the conversion formula; a candidate straight line selection unit (500) to select straight lines that passes inside of a triangles in a converted coordinate system; and a depth value calculation unit (600) to calculate a distance from a star point of a straight line to the triangle in the converted coordinate system. [Reference numerals] (100) Normal plane definition unit; (200) Direction vector classification unit; (300) Normal plane conversion unit; (400) Coordinate system conversion unit; (500) Candidate straight line selection unit; (600) Depth value calculation unit; (AA) Conversion formula; (BB) Conversion coordinate system; (CC) Candidate straight line; (DD) Depth information
Abstract:
PURPOSE: A camera parameter estimating and object extracting system for automatic restoration of a three-dimensional object is provided to extract an object while extracting an object area from an image by making fewer calculations. CONSTITUTION: A background modeling unit (100) forms the lattice background of a photographing object and obtains an image including the background. The background modeling unit extracts feature points of the image and performs background modeling based on the extracted feature points. A camera parameter estimating unit (200) automatically estimates a camera parameter from the extracted feature points. An object area extracting unit (300) extracts an object in the image using the background modeling information. [Reference numerals] (100) Background modeling unit; (200) Camera parameter estimating unit; (300) Object area extracting unit; (AA) Feature point; (BB) Background region information; (CC) Camera parameter; (DD) Object area
Abstract:
본발명은실감파노라마영상생성을위한 3D 데이터포맷생성/소비장치및 방법에관한것으로서, 복수개의카메라를통해촬영된영상들간의매칭포인트를찾고, 상기촬영된영상간에깊이값, 텍스쳐값, 및객체분리정보중에서적어도하나를영상정보로서추출하는영상전처리부, 상기추출된영상정보를실감영상생성에이용하기위해 3D 데이터를구조화하는영상정보구조화부, 상기구조화된 3D 데이터의포맷정보를 DB로저장하는 3D 데이터포맷 DB부, 상기저장된 3D 데이터의포맷정보를이용하여실감파노라마를생성하는실감영상생성부, 및상기생성된실감파노라마를렌더링처리하는실감영상렌더링부를포함할수 있다.
Abstract:
PURPOSE: An image compression system for detecting a license plate and a method thereof are provided to minimize the loss of a license plate area which is an interest area in images of a vehicle by compressing the license plate area including important information with a method which is different from other areas. CONSTITUTION: A license plate area extraction unit (100) extracts a license plate area from images of a vehicle including the license plate area. A license plate area information prediction unit (200) predicts information of the extracted license plate area. A license plate area compression unit (300) compresses the predicted information of the license plate area, the predicted license plate area, and a residual signal with information of a real license plate area. A non license plate area compression unit (400) compresses a non license plate area of the images of the vehicle. [Reference numerals] (100) License plate area extraction unit; (200) License plate area information prediction unit; (210) Area information estimation unit; (220) Residual signal image generating module; (300) License plate area compression unit; (400) Non license plate area compression unit; (AA) Vehicle image; (BB) Predicted bit image; (CC) Residual signal image
Abstract:
PURPOSE: A three-dimensional object's depth information representing and recording system is provided to form a three dimensional surface considering the characteristics of a display device showing a three-dimensional object by representing and recording the depth information of the three-dimensional object by the line segment. CONSTITUTION: A three dimensional surface forming unit (100) defines a three-dimensional surface on which a three dimensional object is projected. A depth information projecting unit (200) calculates the depth from the three-dimensional surface to the three-dimensional object. The depth information projecting unit calculates the distance from a intersecting point between a straight line which crosses at one point defined on the three-dimensional surface and has the same direction vector as a normal vector of the point on the three-dimensional surface and the three dimensional surface to a intersecting point between the straight line and the thee-dimensional object. A depth information recording unit (300) records the calculated depth information on the three-dimensional surface. [Reference numerals] (100) Three dimensional surface forming unit; (AA) Depth information projecting unit; (BB) Depth information recording unit