Complexity-aware adaptive spatial pre-processing for ROI scalable video coding with dynamic transition region

Abstract

We present a complexity-aware adaptive spatial pre- processing (pre-filtering) scheme for the efficient Region-of-Interest (ROI) Scalable Video Coding (SVC). According to the proposed approach, we adaptively vary various parameters of the SVC pre-filters, such as standard deviations, a number of filters for the dynamic pre-processing of a transition region between the ROI and background, etc., thereby enabling to dynamically adjust the desired SVC settings. In addition, our adaptive spatial pre-filtering system is based on an SVC computational complexity-rate-distortion (C-R-D) analysis, thereby adding a complexity dimension to the conventional Region-of-Interest SVC R-D analysis. As a result, the ROI SVC visual presentation quality is significantly improved, which can be especially useful for various resource-limited devices, such as mobile devices. The performance of the presented adaptive spatial ROI SVC pre-processing scheme is evaluated and tested in detail, further comparing it to the Joint Scalable Video Model reference software (JSVM 9.19) and demonstrating significant improvements.