Hardware module for low-resource and real-time stereo vision engine using semi-global matching approach

Stereo matching systems that generate dense, accurate, robust and real-time disparity maps are quite attractive for a variety of applications. Most of the existing stereo matching systems that fulfill to all of these requirements adopt the Semi-Global Matching (SOM) technique. This work proposes a scalable architecture based on a systolic array, fully pipeline. The design builds on a combination of multilevel parallelisms such as image line processing (two-dimensional processing) and disparity. The implementation of the SOM technique combined with the gradiente filter sobel as a pre-processing step and absolute differences as a local similarity method is a very robust approach to both high-quality images from the latest version of the Middlebury image database (22.7% of bad pixels) and low-quality images from our stereo camera system. This whole system was implemented in the FPGA platform Cyclone IV generating images of disparity in HD resolution (1024×768 pixel), with the range of 128 levels of disparities values and using four directions of paths for the SOM method. The proposed approach provided an operating frequency of 100Mhz, delivering images at a rate of 127 frames per second, using 70% of its resource in logical elements for processing and 63% of internal memory for intermediate data storage. So the proposed architecture fills the demands of real world applications regarding frame rate, depth resolution, low resource usage and accuracy.