The present disclosure relates to an image processing apparatus and an image processing method capable of suppressing a deterioration in image quality. The image processing apparatus includes a threshold value setting unit which sets a threshold value for identifying a feature of neighboring pixels of a current block in an intra prediction process in encoding of image data according to a bit depth of the image data and a filtering processing unit which performs a filtering process on the neighboring pixels by using a filter according to the feature of the neighboring pixels identified by using the threshold value set by the threshold value setting unit. The present disclosure may be applied to, for example, an image processing apparatus.

The present disclosure relates to an image processing apparatus and an image processing method capable of suppressing a deterioration in image quality. The image processing apparatus includes a threshold value setting unit which sets a threshold value for identifying a feature of neighboring pixels of a current block in an intra prediction process in encoding of image data according to a bit depth of the image data and a filtering processing unit which performs a filtering process on the neighboring pixels by using a filter according to the feature of the neighboring pixels identified by using the threshold value set by the threshold value setting unit. The present disclosure may be applied to, for example, an image processing apparatus.

In various embodiments, a training application trains a machine learning model to preprocess images. In operation, the training application computes a chroma sampling factor based on a downscaling factor and a chroma subsampling ratio. The training application executes a machine learning model that is associated with the chroma sampling factor on data that corresponds to both an image and a first chroma component to generate preprocessed data corresponding to the first chroma component. Based on the preprocessed data, the training application updates at least one parameter of the machine learning model to generate a trained machine learning model that is associated with the first chroma component.

In various embodiments, a training application trains a machine learning model to preprocess images. In operation, the training application computes a chroma sampling factor based on a downscaling factor and a chroma subsampling ratio. The training application executes a machine learning model that is associated with the chroma sampling factor on data that corresponds to both an image and a first chroma component to generate preprocessed data corresponding to the first chroma component. Based on the preprocessed data, the training application updates at least one parameter of the machine learning model to generate a trained machine learning model that is associated with the first chroma component.

Aspects of the disclosure provide methods and apparatus for video decoding. Processing circuitry of the apparatus decodes coded information for a reconstructed sample of a current component in a current picture from a coded video bitstream. The coded information indicates a sample offset filter to be applied to the reconstructed sample. A filter shape of the sample offset filter is one of a plurality of filter shapes. Each of the plurality of filter shapes includes first reconstructed samples of a first component in the current picture. A filtered sample value of the reconstructed sample is determined based on the first reconstructed samples in the filter shape. The sample offset filter is an in-loop filter by which the output value is applied to the reconstructed sample as an offset to filter out coding artifacts while retaining details of the current component in the current picture.

Aspects of the disclosure provide methods and apparatus for video decoding. Processing circuitry of the apparatus decodes coded information for a reconstructed sample of a current component in a current picture from a coded video bitstream. The coded information indicates a sample offset filter to be applied to the reconstructed sample. A filter shape of the sample offset filter is one of a plurality of filter shapes. Each of the plurality of filter shapes includes first reconstructed samples of a first component in the current picture. A filtered sample value of the reconstructed sample is determined based on the first reconstructed samples in the filter shape. The sample offset filter is an in-loop filter by which the output value is applied to the reconstructed sample as an offset to filter out coding artifacts while retaining details of the current component in the current picture.

An encoding device evaluates a plurality of processing and/or post-processing algorithms and/or methods to be applied to a video stream, and signals a selected method, algorithm, class or category of methods/algorithms either in an encoded bitstream or as side information related to the encoded bitstream. A decoding device or post-processor utilizes the signaled algorithm or selects an algorithm/method based on the signaled method or algorithm. The selection is based, for example, on availability of the algorithm/method at the decoder/post-processor and/or cost of implementation. The video stream may comprise, for example, downsampled multiplexed stereoscopic images and the selected algorithm may include any of upconversion and/or error correction techniques that contribute to a restoration of the downsampled images.

An encoding device evaluates a plurality of processing and/or post-processing algorithms and/or methods to be applied to a video stream, and signals a selected method, algorithm, class or category of methods/algorithms either in an encoded bitstream or as side information related to the encoded bitstream. A decoding device or post-processor utilizes the signaled algorithm or selects an algorithm/method based on the signaled method or algorithm. The selection is based, for example, on availability of the algorithm/method at the decoder/post-processor and/or cost of implementation. The video stream may comprise, for example, downsampled multiplexed stereoscopic images and the selected algorithm may include any of upconversion and/or error correction techniques that contribute to a restoration of the downsampled images.

A method and apparatus for reducing artifacts in a compressed image using a neural-network based deblocking filter. The method may include receiving at least one reconstructed image, wherein each reconstructed image comprises one or more reconstructed blocks and extracting boundary areas associated with boundaries of the one or more reconstructed blocks in the at least one reconstructed image. The extracted boundary areas may be input in a trained deblocking model to generate boundary areas having reduced artifacts and the trained deblocking mode is trained on training data based on estimated compression by a neural image compression (NIC) network. The edge areas associated with the generated boundary areas may be removed; and at least one reconstructed image with reduced artifacts may be generated based on the generated boundary areas.

A method and apparatus for reducing artifacts in a compressed image using a neural-network based deblocking filter. The method may include receiving at least one reconstructed image, wherein each reconstructed image comprises one or more reconstructed blocks and extracting boundary areas associated with boundaries of the one or more reconstructed blocks in the at least one reconstructed image. The extracted boundary areas may be input in a trained deblocking model to generate boundary areas having reduced artifacts and the trained deblocking mode is trained on training data based on estimated compression by a neural image compression (NIC) network. The edge areas associated with the generated boundary areas may be removed; and at least one reconstructed image with reduced artifacts may be generated based on the generated boundary areas.