摘要:
By using parallax compensation which performs prediction by using parallax between video images, the video images are encoded as a single video image. Reference parallax for a target image to be encoded is set, wherein the reference parallax is estimated using a reference image; area division in an image frame is set; parallax displacement for each divided area is set, wherein the parallax displacement is the difference between the reference parallax and parallax for the parallax compensation; data of the area division is encoded; and data for indicating the parallax displacement is encoded. During decoding, reference parallax for a target image to be decoded is set, wherein it is estimated using a reference image; data for indicating area division, which is included in encoded data, is decoded; and data of parallax displacement, which is included in the encoded data, is decoded for each area indicated by the area division data.
摘要:
A video encoding apparatus used in encoding of a multi-viewpoint image. The apparatus generates a synthetic image for a camera used for obtaining an encoding target image, by using an already-encoded reference camera image having a viewpoint different from the viewpoint of the camera used for obtaining the encoding target image, and disparity information between the reference camera image and the encoding target image, thereby encoding the encoding target image. A predicted image for a differential image between an input image of an encoding target area to be encoded and the synthetic image generated therefor is generated, and a predicted image for the encoding target area, which is represented by the sum of the predicted differential image and the synthetic image for the encoding target area, is generated. A prediction residual represented by a difference between the predicted image for the encoding target area and the encoding target image of the encoding target area is encoded.
摘要:
An image generation method for generating image information of a color signal Y of an image A by using a color signal X of image A, and color signal X and color signal Y of an image B. The presence or absence of a point in color signal X of image B corresponding to each pixel position of color signal X of image A, and the position of the relevant corresponding point are estimated. To each estimated pixel position in color signal Y of image A, image information of the corresponding position in the second color signal Y of image B is assigned. Color signal Y at a pixel position in image A for which it is estimated that there is no corresponding point is generated by using the image information of color signal Y assigned to pixels having a corresponding point.
摘要:
An image encoding method includes determining and encoding global parallax data which is probably correct parallax data in consideration of the Epipolar geometry constraint between a camera of a standard viewpoint, which is selected from the entire multi-viewpoint images, and images obtained by all the other viewpoints; generating base parallax data for each camera as a viewpoint other than the standard viewpoint, where the base parallax data is probably correct parallax data in consideration of the Epipolar geometry constraint between the image of the relevant camera and the images of all the other cameras based on the global parallax data and the camera parameters; determining and encoding correction parallax data used for correcting the base parallax data, so as to indicate parallax data between the image of the relevant camera and an already-encoded reference viewpoint image used for parallax compensation; and encoding the image of the relevant camera by using parallax data obtained by correcting the base parallax data by means of the correction parallax data.
摘要:
In order to make it possible to obtain the correct decoded image even in the case of not decoding a particular frame of the encoded data and improve the coding efficiency, the predicted image production unit 103 selects the image data from the image data of a plurality of frames in the reference image memory 107 which are encoded in the past, of the i-th (1≦i≦j) category, for the current frame which is classified as the j-th category by the image classifying unit 102, and produces the predicted image. The difference encoding unit 104 encodes a difference between the image data of the current frame and the predicted image. Also, the current category encoding unit 106 encodes the category number of the current frame, and the reference image specifying data encoding unit 105 encodes the reference image specifying data which specifies the image data selected from the reference image memory 107.
摘要翻译:为了即使在不对编码数据的特定帧进行解码并提高编码效率的情况下也可以获得正确的解码图像,预测图像生成单元103从多个图像数据的图像数据中选择图像数据 对于由图像分类单元102分类为第j个类别的当前帧,过去编码的参考图像存储器107中的第i(1 <= i <= j)个类别的帧, 并产生预测图像。 差编码单元104对当前帧的图像数据和预测图像之间的差进行编码。 此外,当前类别编码单元106对当前帧的类别编号进行编码,并且参考图像指定数据编码单元105对指定从参考图像存储器107中选择的图像数据的参考图像指定数据进行编码。
摘要:
A video encoding method for generating, based on disparity information between an already-encoded reference camera image and an encoding target camera image corresponding to the reference camera image, a disparity-compensated image by means of inter-camera image prediction; and encoding a differential image between the encoding target camera image and the disparity-compensated image. The method includes selecting, for each predetermined section unit on the differential image, one of a decoded differential image group obtained by decoding a differential image between the already-encoded camera image and the disparity-compensated image and a decoded camera image group obtained by decoding the already-encoded camera image, by determining whether or not there is a disparity-compensated image at the corresponding position, that is, whether or not the corresponding pixel in the disparity-compensated image has an effective value.
摘要:
A method for encoding an image using an intraframe prediction is provided which includes selecting a gradient of a pixel value that is indicated by an image signal to be predicted among a plurality of gradient candidates, generating a predicted signal by applying a gradient in accordance with the distance from a prediction reference pixel based on the gradient, intraframe-encoding the image signal to be predicted based on the predicted signal, and encoding information indicating the size of the selected gradient. Alternatively, the method includes estimating the gradient of a pixel value that is indicated by an image signal to be predicted based on an image signal which has already been encoded, generating a predicted signal by applying a gradient in accordance with the distance from a prediction reference pixel based on the gradient, and intraframe-encoding the image signal to be predicted based on the predicted signal.
摘要:
In scalable video encoding, incidence rates of combinations of optimum prediction modes to be selected for spatially corresponding blocks of an upper layer and a lower layer are determined based on an optimum prediction mode that was selected in a conventional encoding, and then a correspondence table that describes relationships therebetween is created. Subsequently, the combinations of the selected optimum prediction modes described in the correspondence table are narrowed down based on the value of the incidence rate so as to create prediction mode correspondence information that describes the combinations of the optimum prediction mode narrowed down. When encoding the upper layer block, the prediction mode search candidate that is to be searched in the encoding is decided by referring to the prediction mode correspondence information using as a key the optimum prediction mode selected in the encoding of the spatially corresponding lower layer block and thereby reducing the number of the prediction mode search candidates.
摘要:
An image generation method for generating image information of an image C by using an image A and an image B having a bit depth higher than that of image A. Image C having the same bit depth as image B is generated by increasing the bit depth of image A by means of tone mapping; presence or absence of a point in image B corresponding to each pixel position of image C and the position of the relevant corresponding point are estimated; and to each pixel position in image C for which it is estimated that there is a corresponding point, image information of the corresponding position in image B is assigned. It is possible to generate image information at each pixel position in image C for which it is estimated in the corresponding point estimation that there is no corresponding point, by using the image information assigned according to an estimation result that there is a corresponding point.
摘要:
A direction is detected for each block in which a pixel value is changed which is represented by an edge that indicates a direction of change in pixel value in each block, a direction in which a deblocking filter is to be applied to a block boundary is determined based on a direction of an edge detected for a block to be processed which includes the block boundary subject to deblocking and on a direction of an edge detected for a block contacting the block to be processed, and the deblocking filter is applied to the block boundary in accordance with the determined direction.