Abstract:
본 발명의 확률적 렌더링에 따른 이미지의 잡음 제거 방법은 확률적 렌더링을 통해 입력 잡음 이미지를 획득하는 단계; 상기 입력 잡음 이미지의 대상과 동일한 대상에 대해 보조 이미지를 획득하는 단계; 상기 입력 잡음 이미지와 상기 보조 이미지를 입력받아, 상기 보조 이미지를 상기 입력 잡음 이미지의 추정 값으로 사용하여 결합 양방향 필터링함으로써 결과 이미지를 획득하는 단계를 포함한다.
Abstract:
PURPOSE: A 3D volume-plan coupled image generating method is provided to obtain an image in which a planer image of a body part is coupled with related volumetric portions, thereby providing images more precise and easier to understand in the medical fields. CONSTITUTION: A method for generating 3D volume-plan coupled images includes a volume rendering step of shearing an original image data with respect to an object by the shear-warp algorithm to obtain volume data on the sheared plan and generating an intermediate volume image formed on an intermediate image plan by mixing(S11), an MPR(Multi-Planer Reconstruction) image extracting step of generating topographical images with respect to a selected portion by MPR and carrying out affine transform proper to the intermediate image plan, an image coupling step of coupling the intermediate volume image with the MPR image(S12), and a volume-plan coupled image generating step of warping the intermediate volume image-MPR coupled image for generating a final volume-plan coupled image with respect to the object(S28,S29).
Abstract:
PURPOSE: An image search based image watermarking system, a method thereof, and computer readable medium including a command are provided to reduce the number of images for searching for a watermark pattern by filtering data through image search. CONSTITUTION: An image similarity calculating unit(100) calculates image similarity between images in first database and stores the image in second database. A watermark pattern extracting unit(200) extracts a watermark pattern in the second database. An image extracting unit(300) calculates similarity between a watermark pattern of an inputted image and a watermark pattern of an extracted image.
Abstract:
본 발명은 3차원 의료영상 데이터의 가시화방법에 관한 것으로서, 더 상세하게는 부 표본 3차원 데이터로부터 저해상도 영상을 구하고 저해상도 영상 또는 깊이정보와 3차원 데이터를 이용, 고화질의 영상을 구현하여 의료영상의 3차원 데이터를 가시화하고 실시간 처리를 할 수 있는 3차원 의료영상 데이터의 가시화방법에 관한 것이다. 본 발명은 3차원 데이터를 2x2x2의 크기 블록으로 평균하여 부 표본화된 데이터에서 저해상도 영상 또는 깊이 정보를 구하고 2차원으로 보간하여 저해상도 영상을 구하는 제1단계와, 상기 제1단계의 저해상도 영상과 원 데이터를 이용하여 고해상도 영상을 구하는 제2단계로 구성된다. 본 발명에 의하면 고해상도의 MIP 및 VR 영상을 얻는 데 필요한 계산시간을 절약하여 화질을 가변하면서 빠르게 고화질 영상을 볼 수 있는 효과가 있다.
Abstract:
PURPOSE: A method for visualizing three-dimensional medical video data is provided to perform MIP(maximum Intensity Projection) and VR(Volume Rendering) in low resolution, and to implement a hierarchical MIP and VR video to be processed fast, so as to reduce a time, implement high-resolution video and help medical treatment of doctors. CONSTITUTION: A low-resolution MIP(14a) is calculated from sub-sampled data(10) for high-speed processing, and is interpolated(16) in two-dimensional to calculate low-resolution video(20). High-resolution MIP video(22) is calculated using the low-resolution video(20) and full original data(12). The sub-sampling in the step of S10 is obtained by averaging three-dimensional data to a 2x2x2 size block, wherein the 2x2x2 size block is related with internal values as well as adjacent voxels(volume elements). Calculating the low-resolution MIP(14a) from the sub-sampled data(10) enables high-speed processing and is appropriate for viewing a concerned area. Values of the low-resolution MIP(14a) and the sub-sampled data(10) are compared, to obtain entire resolution data corresponding to the value having higher relation.
Abstract:
PURPOSE: A noise removal method of an image according to the probabilistic rendering is provided to preserve the image feature of an input noise image, which is obtained through the probabilistic rendering, and effectively remove the noise; thereby improving the image quality. CONSTITUTION: An input noise image (110) is obtained through the probabilistic rendering (100). As a dispersed ray tracing (200) using the flash shading is performed to an object, which is equal to an object of the input noise image, the support images (210,220) are obtained. After the input noise image and the support images are inputted, a result image (310) is obtained by a joint bidirectional filtering (300) of the input noise image and the support images. The support images are used as an estimate value of the input noise image. [Reference numerals] (100) Probabilistic rendering; (110) Input noise image; (200) Dispersed ray tracing; (210) Virtual flash image; (220) Depth image; (300) Joint bidirectional filtering; (310) Result image
Abstract:
PURPOSE: A 3D volume-plan coupled image generating method is provided to obtain an image in which a planer image of a body part is coupled with related volumetric portions, thereby providing images more precise and easier to understand in the medical fields. CONSTITUTION: A method for generating 3D volume-plan coupled images includes a volume rendering step of shearing an original image data with respect to an object by the shear-warp algorithm to obtain volume data on the sheared plan and generating an intermediate volume image formed on an intermediate image plan by mixing(S11), an MPR(Multi-Planer Reconstruction) image extracting step of generating topographical images with respect to a selected portion by MPR and carrying out affine transform proper to the intermediate image plan, an image coupling step of coupling the intermediate volume image with the MPR image(S12), and a volume-plan coupled image generating step of warping the intermediate volume image-MPR coupled image for generating a final volume-plan coupled image with respect to the object(S28,S29).
Abstract:
본 발명의 이미지 검색 기반의 이미지 워터마킹 시스템은, 제1 데이터베이스에 저장된 다수의 이미지와 입력된 이미지 사이의 이미지 유사도를 계산하여, 상기 이미지 유사도가 제1 설정기준보다 높은 이미지를 제2 데이터베이스에 저장하는 이미지 유사도 계산부; 상기 이미지 유사도 계산부의 제2 데이터베이스에 저장된 이미지에서 워터마크 패턴을 추출하는 워터마크 패턴 추출부; 및 상기 이미지 유사도 계산부에 입력된 이미지의 워터마크 패턴과 상기 워터마크 패턴 추출부에서 추출된 워터마크 패턴 사이의 워터마크 패턴 유사도를 계산하여, 상기 워터마크 패턴 유사도가 제2 설정기준보다 높은 이미지를 추출하는 이미지 추출부;를 포함한다.