Clustering-Based Linear Mappings Learning For Quantization Noise Removal



Clustering-Based Linear Mappings Learning For Quantization Noise Removal

Clustering-Based Linear Mappings Learning For Quantization Noise Removal
Research Paper / IEEE 2016 International Conference on Image Processing (ICIP) / Sep 2016 / Image Processing, Video Coding

This paper describes a novel scheme to reduce the quantization noise of compressed videos and improve the overall coding performances. The proposed scheme first consists in clustering noisy patches of the compressed sequence. Then, at the encoder side, linear mappings are learned for each cluster between the noisy patches and the corresponding source patches. The linear mappings are then transmitted to the decoder where they can be applied to perform de-noising. The method has been tested with the HEVC standard, leading to a bitrate saving of up to 9.63%.