Abstract:
본 발명은 움직임 추정방법 및 움직임 추정장치, 이를 이용한 부호화 방법 및 비디오 인코더, 이를 이용한 복호화 방법 및 비디오 디코더에 관한 것이다. 본 발명에 따른 움직임 추정방법은, 현재 영상 블록의 고주파 및 저주파 성분을 추정하는 방법에 있어서, (a) 현재 영상 블록과 현재 영상 블록의 레퍼런스 후보 블록들의 픽셀값에 현재 영상 블록과 현재 영상 블록의 레퍼런스 블록들의 평균값을 뺀 값을 이용하여 유사도를 분석하여, 현재 영상 블록과 고주파 성분 유사도가 가장 높은 레퍼런스 블록을 선택하고, 선택된 레퍼런스 블록의 고주파 성분을 현재 영상 블록의 고주파 성분으로 추정하는 단계, (b) 현재 영상 블록의 인접한 이웃 픽셀들과 선택된 레퍼런스 블록의 인접한 이웃 픽셀들의 평균값을 뺀 값을 이용하여, 현재 영상 블록의 저주파 성분을 추정하는 단계, (c) 추정된 저주파 성분을 선택된 레퍼런스 블록에 적용하여 움직임 보상된 현재 영상 블록의 예측 블록을 생성하는 단계를 포함한다. 움직임 추정, 고주파 성분, 저주파 성분, 레퍼런스 블록
Abstract:
PURPOSE: A method and device for modeling a virtual channel for dispersion video compression technique are provided to perform modeling of a virtual channel to coincide a Laplacian distribtuiosn of a channel parameter of a virtual channel with a Laplacian distribution. CONSTITUTION: A difference frame is generated(S100). A 4x4 DCT(Discrete Cosine Transform) conversion about the difference frame is operated(S110). Samples in predetermined reliable section among samples are selected(S120). A sample dispersion based on reliable section area is calculated(S130). A virtual channel parameter is calculated(S140). By modeling a virtual channel by using the virtual channel parameter, optimum probability distribution is obtainer(S150).
Abstract:
PURPOSE: A motion estimating method and a motion estimating apparatus, an encoding method and a video encoder for using the same, a decoding method and a video decoder for using the same are provided to independently estimate the low frequency components of a current image block by using adjacent pixels of current and reference blocks in order to accurately estimate motion. CONSTITUTION: A video encoder adopting a motion estimation method comprises a high frequency estimating unit(110), a low frequency estimation unit(120), a motion compensating unit(130), a conversion unit(140) and an entropy encoding unit(150). The conversion unit generates a residual block by deducting an estimation block for which motion is compensated, converts the information of the residual block into a frequency, and quantizes the residual block. The entropy encoding unit encodes the reproduction information for reproducing the motion of a current image block and the information of the quantized residual block.
Abstract:
PURPOSE: A video signal processing method and a device are provided to improve compression efficiency by using nine kinds intra prediction. CONSTITUTION: A video signal receiving unit receives video signal including prediction mode information, interpolating information and a residual of current block. An interpolating pixel recovering unit(640) restores a interpolating pixel by using the interpolating information and a neighboring block. A present block restoration unit(650) restores the present block by using the interpolating pixel, the prediction mode information and the residual. The interpolating pixel recovering unit uses the interpolating information generated based on location of the present block.
Abstract:
PURPOSE: A higher order generalized series parallel imaging method and a sampling method are provided to reduce an image obtaining time by combining a parallel magnetic resonance imaging method and a higher order generalized imaging method. CONSTITUTION: An input image is sampled in a k-space. A first re-composition image is obtained by applying a parallel magnetic resonance imaging method to the data obtained by the sampling. A second re-composition image is obtained by applying a higher order generalized imaging method to the first re-composition image. The parallel magnetic resonance imaging method is a SPACE RIP(Sensitivity Profiles From Array of Coils for Encoding and Reconstruction in Parallel), SENCE, and PILS or GRAPPA.
Abstract:
본 발명은 고속모드선택방법 및 그 기록매체에 관한 것이다. 이러한 본 발명의 일 실시 예에 따른 고속모드선택 방법은 (a) 매크로블록 내의 복수의 모드(mode)들에 대한 각각의 움직임 비용 값(J motion )들을 계산하는 단계, (b) 계산된 움직임 비용 값들을 기초로 하여, 복수의 모드들 중 가장 작은 움직임 비용 값을 갖는 모드를 포함하는 일부 모드들에 대한 각각의 율-왜곡 비용 값(Rate Distortion cost, RD cost)들을 계산하는 단계 및 (c) 계산된 율-왜곡 비용 값들 중 가장 작은 율-왜곡 비용 값(RDcost1)을 갖는 모드를 최적모드(best mode)로 선택하는 단계를 포함한다.
Abstract:
본 발명은 위상 보정 방법에 관한 것으로, 더욱 상세하게는, 비선형 위상 보정 방법에 관한 것이다. 본 발명에 의한 비선형 위상 보정 방법은 그레디언트 에코(gradient echo) EPI(Echo Planar Imaging)에 의해 영상 정보를 획득하는 단계, 스핀 에코(spin echo) EPI에 의해 기준 정보를 획득하는 단계 및 상기 기준 정보를 이용하여 상기 영상 정보를 보정하는 단계를 포함한다. 자기 공명, 그레디언트 에코, 스핀 에코, 비선형 위상 보정
Abstract:
A non-linear phase correction method is provided to acquire a stable image by correcting a phase according to reference information through spin echo EPI(Echo Planar Imaging). A non-linear phase correction method includes the steps of: obtaining image information through gradient echo EPI; obtaining reference information through spin echo EPI; and correcting the image information based on the reference information. The correction is performed by deducting a phase value of the reference information from a phase value of the image information. The phase value of the image information is a phase of information obtained through 1 dimensional Fourier Transform of the image information. The phase value of the reference information is a phase of information obtained through 1 dimensional Fourier Transform and projection of the reference information.
Abstract:
A method of reducing a ghost artifact in an EPI by using sensitivity encoding and a computer readable record medium with a program for executing the method are provided to remove an aliasing signal by using the sensitivity encoding. An initial image is obtained by using a multiple channel coil of an EPI(Echo Planar Image), in which an odd image and an even image are alternatively obtained from left and right parts of a path obtaining a signal of a frequency region. Only even image is collected from the initial image, and then is Fourier-transformed to obtain an even intermediate image. Only odd image is collected from the initial image, and then is Fourier-transformed to obtain an odd intermediate image. The even intermediate image and the odd intermediate image are sensitivity-encoded to obtain an even image and an odd image. A final image is generated by taking an average of the even image and odd image.
Abstract:
PURPOSE: A method for transforming the size of an image at a block discrete cosine transform domain is provided to randomly transform the size of the image and remarkably reduce the amount of calculation by simplifying window functions. CONSTITUTION: In the case of downsampling, 8xM samples are made by inserting M-1 zeros in the back of each sample of a DCT(Discrete Cosine Transform) block formed of 8 samples. The adjacent 8xM samples are symmetrically convoluted with DCT coefficients of M window functions. Upper 8 samples of the M results of the symmetrical convolution are added with each other to make 8-sample DCT. In the case of upsampling, 8xL samples are made by inserting (L-1)x8 zeros after each sample of a DCT block formed of 8 samples. 8xL samples are symmetrically convoluted with DCT coefficients of L window functions. Every L sample is collected from the 8xL sample result value of the symmetrical convolution to make L 8-sample DCT block.