Abstract:
스테레오 매칭 결과를 개선하는 장치가 개시된다. 이 스테레오 매칭 결과를 개선하는 장치는 다수의 전처리 조건에 따라 전처리된 두 영상 간의 양안 시차를 이용하여 양안 시차 영상을 출력하는 스테레오 카메라부와, 상기 양안 시차 영상에 대해 이산 코사인 변환(Discrete Cosine Transform: DCT)하여 DCT 계수를 생성하는 이산 코사인 변환부와, 상기 생성된 DCT 계수를 입력받고, 상기 DCT 계수들 중 상기 줄무늬 패턴이 포함된 AC 계수들을 이용하여 화면에 분포한 줄무늬 양을 평가하는 줄무늬 평가부와, 상기 다수의 전처리 조건 중 상기 평가된 줄무늬 양 중 가장 낮은 줄무늬 양에 대응하는 전처리 조건을 최적의 조건을 평가하는 조건 평가부 및 상기 이산 코사인 변환부에 의해 획득된 상기 DCT 계수 중 소정의 AC 계수들을 변경하고, 변경된 상기 DCT 계수를 역 이산 코산인 변환하여 상기 줄무늬가 제거된 양안 시차 영상을 생성하는 줄무늬 제거부를 포함한다.
Abstract:
PURPOSE: A device and method for correcting a projector image are provided to find distance information about a projection plane using a distance image camera in order to detect the surface shape of the plane and in order to identify surface conditions such as colors or materials and the condition of a peripheral lighting, thereby effectively correcting an output target image. CONSTITUTION: A device for correcting a projector image includes a distance image calculating part (102), a surface shape detecting part (103), a surface condition detecting part, a lighting condition detecting part, and an output image correcting part (106). The distance image calculating part outputs the calculation results of distance images from distance image cameras (101a-101n). The surface shape detecting part recognizes the surface shape of a projection plane using the results, and obtains geometric information. The surface condition detecting part analyzes the colors and materials of an independent projection circumstance image configured with images from actual image cameras in order to divide a screen into areas and in order to analyze the colors and materials for the respective areas. The lighting condition detecting part obtains the lighting information of a circumstance. The output image correcting part corrects an output image using the information. [Reference numerals] (101a) Camera #1; (101n) Camera #n; (102) Distance image calculating part; (103,104,105) Surface shape detecting part; (106) Output image correcting part; (AA) Output image; (BB) Image output device
Abstract:
PURPOSE: An apparatus for improving a stereo matching result, a method for improve the stereo matching result using the apparatus, and a system for receiving the stereo matching result are provided to acquire a binocular parallax image in which a horizontal stripe pattern is removed in optimal conditions. CONSTITUTION: A stereo camera unit(110) outputs a binocular parallax image between two images which are pre-processed. A DCT(Discrete Cosine Transform) unit(120) performs the DCT of the binocular parallax image. A stripe evaluation unit(130) evaluates the amount of stripes on a screen. A condition evaluation unit(140) evaluates pre-processing conditions corresponding to the smallest stripe amount of the evaluated stripe amount as optimal conditions. According to the optimal condition, a stripe removing unit(150) eliminates the stripe.
Abstract:
A sight stabilization apparatus for a mobile robot is provided to perceive vibration of the sight of a robot camera by analyzing vibration or image of the robot camera and controlling motion of the sight of the robot camera by the compensation value calculated in accordance with the vibration of the sight. A sight stabilization apparatus for a mobile robot includes a camera vibration sensing unit(11), a travel pattern analysis unit(12), a sight control unit(13), a sight adjustment unit(16). The camera vibration sensing unit senses the vibration of a camera generated during travel of the mobile robot. The travel pattern analysis unit analyzes the travel pattern of the mobile robot in accordance with the shape, travel method and travel pattern. The sight control unit calculates a vibration compensation value for compensating for the sensed vibration of the camera on the basis of the analyzed travel pattern. The sight adjustment unit adjusts the direction of the sight of the camera in accordance with the vibration compensation value of the sight control unit.
Abstract:
An apparatus for acquiring a 3D depth image is disclosed. The apparatus for acquiring a 3D depth image includes: a pattern scanning unit which scans a plurality of patterned light with different base distances and phases; an image photographing unit which photographs an image using reflected light by the patterned light reflected from an object; and an image processing unit which forms permutation information of patterns, based on the patterns extracted from the photographed image, determines a degree of shifting of the patterns, based on the permutation information of the patterns, to estimate 3D depth information, and acquires a 3D depth image, based on the 3D depth information.
Abstract:
PURPOSE: A stereo image processing unit and a method thereof are provided to analyze one or more of stability of the stereo image, number of feature point and lighting condition of camera images in real-time, thereby creating correct pattern for the stereo matching and projecting the corrected pattern as feedback values to the subject. CONSTITUTION: A method of a stereo image processing unit comprises the following steps: an image processing unit(310) processes images about target objects; a stereo image generator(311) creates the processed images as the stereo images; an unstable area analysis unit(312) analyzes an unstable area in the stereo images in real-time; and a correction pattern generator(313) creates correct pattern which will reflect in the stereo matching based on the analyzed result.
Abstract:
본 발명은 게임 상에서 로봇의 제어 명령으로서 인식되는 사용자 자세에 따라 로봇의 동작을 조절할 수 있는 게임 시스템 및 방법을 제공하는 것으로, 본 발명의 게임 시스템은, 제 1 사용자에게 부착되어 상기 제 1 사용자의 자세를 검출하는 제 1 자세 검출 소자; 및 상기 제 1 자세 검출 소자에 의해 검출되는 신호를 통해 상기 제 1 사용자의 자세를 인식하고, 상기 인식된 제 1 사용자의 자세에 따라 제 1 로봇의 동작을 제어하는 제 1 게임 단말을 포함하는 것을 특징으로 한다. 사용자자세, 로봇, 게임, 자세인식
Abstract:
PURPOSE: An image correction apparatus and an image correction method using the same are provided to correct an image only in the ROI of a user. CONSTITUTION: An image input unit(120) creates a plurality of images. The image input unit performs the pre-processing of the created image. A region extracting unit(149) receives the input of the pre-processed image. The region extracting unit extracts ROI(Region Of Interest). An image correcting unit(164) corrects an image corresponding to the extracted ROI.