Statistics for estimating quantization factors of a coding-unit type (e.g., B-coded or I-coded) pictures are determined from other, possibly different (e.g., P-coded) pictures, or previously coded coding-units. Bit rate and quality relationships between such coding-unit types may be used with the quantization parameters. Estimating bit rate and quality relationships between coding-unit types enables accurate rate control for pictures regardless of their coding-unit type. Bit rate and quality relationships between coding-unit types can be used with multiple rate control models to increase compression. Rate control parameters may be adjusted with statistics generated by a motion estimation and compensation framework. Rate control performance may be controlled in transcoding compressed bit streams.

Rate control model adaptation based on slice dependencies for video coding

Techniques and tools for encoding enhancement layer video with quantization that varies spatially and/or between color channels are presented, along with corresponding decoding techniques and tools. For example, an encoding tool determines whether quantization varies spatially over a picture, and the tool also determines whether quantization varies between color channels in the picture. The tool signals quantization parameters for macroblocks in the picture in an encoded bit stream. In some implementations, to signal the quantization parameters, the tool predicts the quantization parameters, and the quantization parameters are signaled with reference to the predicted quantization parameters. A decoding tool receives the encoded bit stream, predicts the quantization parameters, and uses the signaled information to determine the quantization parameters for the macroblocks of the enhancement layer video. The decoding tool performs inverse quantization that can vary spatially and/or between color channels.

Adaptive quantization for enhancement layer video coding

Clusters of pixels are defined for use in image compression and decompression The image information used to define the clusters may include pixel values at predetermined positions relative to a pixel or related motion vectors, gradients, texture etc During compression of images the image information relative to pixels is examined to determine the cluster to which it belongs Thus pixels can be classified according to the cluster for their image information In an embodiment the definitions of the clusters are selected dynamically, dependent on image content For each cluster a control parameter set is computed for a post-processing operation, such as filter coefficients for filtering or statistical data for locally generating texture The control parameter set is selected dependent on the image content so that, when the post-processing operation is applied to the image after decompression it will improve image quality for the pixels that are classified as belonging to the cluster The compressed image and the control parameter sets are transmitted to a decompression apparatus Upon decompression, the image information that represents the decompressed image is examined to classify pixels according to the clusters and the different control parameter sets for the selected clusters are used to control post-processing at the locations of the pixels

Image compression and decompression

A video encoder identifies a current smooth region of a current picture in a sequence and performs temporal analysis by determining whether a corresponding region in at least one previous and/or future picture is smooth. Based at least in part on the temporal analysis, the encoder adjusts quantization in the current smooth region. An encoder determines a differential quantization interval for a sequence, the interval comprising an interval number. The interval constrains the encoder to skip differential quantization for at least the interval number of predicted pictures after a predicted differentially quantized picture. An encoder analyzes texture in a current picture and sets a smoothness threshold. The encoder compares texture data with the smoothness threshold and adjusts differential quantization for at least part of the current picture based on a finding of at least one smooth region in the current picture according to the smoothness threshold.

Quantization adjustment based on texture level

A video encoder identifies one or more AC coefficients of each of plural blocks in the picture. The encoder identifies a threshold quantization step size such that the identified AC coefficient(s) of each of the plural blocks are nonzero after quantization according to the threshold quantization step size. The threshold quantization step size is such that quantization according to the next higher quantization step size would result in at least one of the identified AC coefficient(s) of at least one of the plural blocks being zero. For example, identifying the threshold quantization step size comprises identifying n top AC coefficients in each of four blocks of a macroblock, determining the smallest AC coefficient among the identified n top AC coefficients of the four blocks, and iteratively evaluating the smallest AC coefficient with respect to candidate quantization step sizes until the threshold quantization step size is identified.

Adjusting quantization to preserve non-zero AC coefficients

Noise is reduced in a video system by applying motion compensated temporal filtering using previously generated motion vectors and adaptive spatial filtering at scene change frames. Various types of noise can be introduced into video prior to compression and transmission. Artifacts arise from recording and signal manipulation, terrestrial or orbital communications, or during decoding. Noise introduced prior to image compression interferes with performance and subsequently impairs system performance. While filtering generally reduces noise in a video image, it can also reduce edge definition leading to loss of focus. Filtering can also tax system throughput, since increased computational complexity often results from filtering schemes. Furthermore, the movement of objects within frames, as defined by groups of pixels, complicates the noise reduction process by adding additional complexity. In addition to improvements made to FIR spatial filtering, the present invention improves on previous filtering techniques by using Infinite Impulse Response (IIR) temporal filtering to reduce noise while maintaining edge definition. It also uses motion vectors previously calculated as part of the first-pass image encoding or alternatively by transcoding to reduce computational complexity for P-frame and B-frame image preprocessing. Single stage P-frame temporal noise filtering and double stage B-frame temporal noise filtering are presented.

Noise reduction pre-processor for digital video using previously generated motion vectors and adaptive spatial filtering

Noise is reduced in a video system by applying motion compensated temporal filtering using previously generated motion vectors and adaptive spatial filtering at scene change frames. Various types of noise can be introduced into video prior to compression and transmission. Artifacts arise from recording and signal manipulation, terrestrial or orbital communications, or during decoding. Noise introduced prior to image compression interferes with performance and subsequently impairs system performance. While filtering generally reduces noise in a video image, it can also reduce edge definition leading to loss of focus. Filtering can also tax system throughput, since increased computational complexity often results from filtering schemes. Furthermore, the movement of objects within frames, as defined by groups of pixels, complicates the noise reduction process by adding additional complexity. In addition to improvements made to FIR spatial filtering, the present invention improves on previous filtering techniques by using Infinite Impulse Response (IIR) temporal filtering to reduce noise while maintaining edge definition. It also uses motion vectors previously calculated as part of the first-pass image encoding or alternatively by trans coding to reduce computational complexity for P-frame and B-frame image preprocessing. Single stage P-frame temporal noise filtering and double stage B-frame temporal noise filtering are presented.

Motion compensated temporal filtering for noise reduction pre-processing of digital video data

Noise is reduced in a video system by applying motion compensated temporal filtering using previously generated motion vectors and adaptive spatial filtering at scene change frames. In addition to improvements made to FIR spatial filtering, the present invention improves on previous filtering techniques by using Infinite Impulse Response (IIR) temporal filtering to reduce noise while maintaining edge definition. It also uses motion vectors previously calculated as part of the first-pass image encoding or alternatively by trans coding to reduce computational complexity for P-frame and B-frame image preprocessing. Single stage P-frame temporal noise filtering and double stage B-frame temporal noise filtering are presented.

Jing Yang Chen

Papers

Noise reduction pre-processor for digital video using previously generated motion vectors and adaptive spatial filtering

Motion compensated temporal filtering for noise reduction pre-processing of digital video data

Motion-adaptive noise reduction pre-processing for digital video signal