Abstract: The present disclosure relates to a method for image patch selection for training a neural network for image quality assessment. The method includes receiving an input image and extracting one or more image patches from the input image. The moment of the extracted image patches is measured. There is a decision to accept or decline the extracted image patches according to the measured moment. Additional image patches are extracted until a minimum number, Nmin, of extracted image patches are accepted. Alternatively, selection criteria are adjusted until the minimum number of extracted image patches are accepted. The selected image patches are input into a neural network with a corresponding image quality value of the input image, and the neural network is trained with the image patches and image quality value. Also provided is a method for image quality assessment using a neural network trained as set forth above.

Abstract: The present disclosure relates to a method for image patch selection for training a neural network for image quality assessment. The method includes receiving an input image and extracting one or more image patches from the input image. The moment of the extracted image patches is measured. There is a decision to accept or decline the extracted image patches according to the measured moment. Additional image patches are extracted until a minimum number, Nmin, of extracted image patches are accepted. Alternatively, selection criteria are adjusted until the minimum number of extracted image patches are accepted. The selected image patches are input into a neural network with a corresponding image quality value of the input image, and the neural network is trained with the image patches and image quality value. Also provided is a method for image quality assessment using a neural network trained as set forth above.

Abstract: To reduce the complexity of rate-distortion cost computation in AVC (MPEG4 Part2/H.264), a bit rate estimation method is used to avoid entropy coding during Intra and Inter mode decision. The bit rate estimation method predicts the number of entropy coded bits based on the quantized transform coefficients, thus avoiding the need to entropy code the quantized transform coefficients at each decision iteration.

Abstract: The present embodiments include methods, systems, and apparatuses for foreground biased depth map refinement in which horizontal gradient of the texture edge in color image is used to guide the shifting of the foreground depth pixels around the large depth discontinuities in order to make the whole texture edge pixels assigned with foreground depth values. In such an embodiment, only background information may be used in hole-filling process. Such embodiments may significantly improve the quality of the synthesized view by avoiding incorrect use of foreground texture information in hole-filling. Additionally, the depth map quality may not be significantly degraded when such methods are used for hole-filling.

Abstract: The present invention relates to method and apparatus for multiview video coding. In particular, the present invention describes a disparity compensated prediction to exploit the inter-view correlation in multiview video coding by providing stretching, compression, and shearing (SCSH) disparity compensation to approximate the actual disparity effects in addition to the translational disparity. A sub-sampled block-matching disparity estimation technique is provided to implement the SCSH disparity compensation which makes use of the interpolated reference frames for subpixel motion and disparity estimation in conventional hybrid video coding structure.

Abstract: In accordance with an example embodiment of the present invention, the present invention provides method and apparatus for motion compensated prediction. Apart from translational motion, zoom motion is taken into account by sampling an interpolated frame with one or more selected sampling rates to generate one or more zoom reference frames; matching a frame with the zoom reference frames; and determining one or more motion data.

Abstract: A motion estimation method uses a search start point and a plurality of search paths extending from the start point. The method finds an image point along the plurality of search paths that has a minimum comparison value with a reference image block. This image point with a minimum comparison value becomes the new search start point, with a plurality of new search paths extending from the new start point. The method then finds the image point along the plurality of new search paths that has a minimum comparison value with the reference image block. The cycle repeats until points on the search paths have comparison values higher than the start point.

Abstract: The present invention relates to a method and apparatus for calculating the Sum of Squared Differences (SSD) between a source block and a reconstructed block of image or video data encoding according to an encoding scheme such as H.264/AVC. In a preferred embodiment, the method computes the SSD by finding the SSD between coefficients of an integer transformed residual block and the corresponding inverse-quantized coefficients. Preferably the inverse quantized coefficients are found with the aid of a look up table. This method may save computing time and processing power compared to calculating the SSD directly from the source and reconstructed blocks. The SSD is related to the distortion caused by encoding and the method may be used in calculating the rate-distortion of a particular encoding mode. One embodiment of the invention encodes a block of data by selecting the encoding mode with the least rate-distortion.

Abstract: In accordance with an example embodiment of the present invention, the present invention provides method and apparatus for motion compensated prediction. Apart from translational motion, zoom motion is taken into account by sampling an interpolated frame with one or more selected sampling rates to generate one or more zoom reference frames; matching a frame with the zoom reference frames; and determining one or more motion data.

Abstract: A motion estimation method uses a search start point and a plurality of search paths extending from the start point. The method finds an image point along the plurality of search paths that has a minimum comparison value with a reference image block. This image point with a minimum comparison value becomes the new search start point, with a plurality of new search paths extending from the new start point. The method then finds the image point along the plurality of new search paths that has a minimum comparison value with the reference image block. The cycle repeats until points on the search paths have comparison values higher than the start point.

Abstract: To reduce the complexity of rate-distortion cost computation in AVC (MPEG4 Pat2/H.264), a bit rate estimation method is used to avoid entropy coding during Intra and Inter mode decision. The bit rate estimation method predicts the number of entropy coded bits based on the quantized transform coefficients, thus avoiding the need to entropy code the quantized transform coefficients at each decision iteration.

Abstract: The present invention relates to a method and apparatus for calculating the Sum of Squared Differences (SSD) between a source block and a reconstructed block of image or video data encoding according to an encoding scheme such as H.264/AVC. In a preferred embodiment, the method computes the SSD by finding the SSD between coefficients of an integer transformed residual block and the corresponding inverse-quantized coefficients. Preferably the inverse quantized coefficients are found with the aid of a look up table. This method may save computing time and processing power compared to calculating the SSD directly from the source and reconstructed blocks. The SSD is related to the distortion caused by encoding and the method may be used in calculating the rate-distortion of a particular encoding mode. One embodiment of the invention encodes a block of data by selecting the encoding mode with the least rate-distortion.