Complexity-aware adaptive preprocessing scheme for region-of-interest spatial scalable video coding

Dan Grois, Ofer Hadar

Research output: Contribution to journalArticlepeer-review

10 Scopus citations


This paper presents a complexity-aware adaptive spatial preprocessing scheme for the efficient scalable video coding (SVC) by employing an adaptive prefilter for each SVC layer. According to the presented scheme, a dynamic transition region is defined between the region of interest (ROI) and background within each video frame, and then various parameters of each prefilter (such as the standard deviation, kernel matrix size, and also a number of filters for the dynamic preprocessing of a transition region between the ROI and the background) are adaptively varied. The presented scheme has proved to be very efficient because it is based on an SVC computational complexity-rate-distortion analysis, thereby adding a complexity dimension to the conventional SVC rate-distortion analysis. As a result, the encoding computational complexity resources are significantly reduced, which is especially useful for portable encoders with limited power resources. The performance of the presented adaptive spatial preprocessing scheme is evaluated and tested in detail from both computational complexity and visual presentation quality points of view, further comparing it with the joint scalable video model reference software (JSVM 9.19) and demonstrating significant improvements.

Original languageEnglish
Article number6727577
Pages (from-to)1025-1039
Number of pages15
JournalIEEE Transactions on Circuits and Systems for Video Technology
Issue number6
StatePublished - 1 Jan 2014


  • ROI scalability
  • Regions-of-interest (ROI) video coding
  • Scalable Video Coding (SVC)
  • high-quality visual presentation
  • image/video coding
  • pre-processing/pre-filtering


Dive into the research topics of 'Complexity-aware adaptive preprocessing scheme for region-of-interest spatial scalable video coding'. Together they form a unique fingerprint.

Cite this