Optimized Belief Propagation Algorithm onto Embedded Multi and Many-Core Systems for Stereo Matching

Jean François Nezan (1) , A. Mercat (1) , P. Delmas , G. Gimelfarb

1 IETR - Institut d'Électronique et des Technologies du numéRique

Stereo matching techniques aim at reconstructing disparity maps from a pair of images. The use of stereo matching techniques in embedded systems is very challenging due to the complexity of the state-of-the-art algorithms. Local stereo matching algorithms are efficiently implemented on GPU and DSP. This paper presents the optimization of the One Dimension Belief Propagation (BP-1D) algorithm. BP-1D is faster than previous algorithms on monocore DSP and its implementation onto multicore DSPs is straightforward. BP-1D implemented on multicore embedded platforms out-performs previous stereo matching implementations reaching real-time performances for resolutions up to 1080p with a 10 Watts power consumption. © 2016 IEEE.

Keywords: Algorithms; Backpropagation; Complex networks; Embedded systems; Image matching; Optimization; Stereo vision; Belief propagation algorithm; Embedded visions; Local stereo matching; Multi-core dsp; Real time performance; State-of-the-art algorithms; Stereo matching; Vision processing; Stereo image processing

Domains: Computer Science [cs]; Networking and Internet Architecture [cs.NI]; Engineering Sciences [physics]; Electronics

Jean François Nezan, A. Mercat, P. Delmas, G. Gimelfarb. Optimized Belief Propagation Algorithm onto Embedded Multi and Many-Core Systems for Stereo Matching. 24th Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2016, Feb 2016, Heraklion, Greece. pp.332--336, ⟨10.1109/PDP.2016.52⟩. ⟨hal-01368103⟩