P. Kumar, K. Palaniappan, A. Mittal, and G. Seetharaman

Lecture Notes in Computer Science (ACIVS), Volume 5807, pgs. 320--332, 2009

parallelization, tracking, fmv, motion, features, dod

PlainText, Bibtex, PDF, URL, DOI, Google Scholar

Abstract

The rapid increase in pixel density and frame rates of mod- ern imaging sensors is accelerating the demand for fine-grained and em- bedded parallelization strategies to achieve real-time implementations for video analysis. The IBM Cell Broadband Engine (BE) processor has an appealing multi-core chip architecture with multiple programming models suitable for accelerating multimedia and vector processing appli- cations. This paper describes two parallel algorithms for blob extraction in video sequences: binary morphological operations and connected com- ponents labeling (CCL), both optimized for the Cell-BE processor. Novel parallelization and explicit instruction level optimization techniques are described for fully exploiting the computational capacity of the Syner- gistic Processing Elements (SPEs) on the Cell processor. Experimental results show significant speedups ranging from a factor of nearly 300 for binary morphology to a factor of 8 for CCL in comparison to equivalent sequential implementations applied to High Definition (HD) video.