WANG Yuan-yuan, LIU Xue-cheng, WU Feng-juan. Implementation of 3D reconstruction algorithm based on binocular vision[J]. 2018, (4): 136-138+141. DOI: 10.13873/J.1000-9787(2018)04-0136-03.
which uses the OV7725 complementary metal oxide semiconductors(CMOS) camera group to collect images
and images are displayed by video graphics array(VGA)after field programmable gate array(FPGA) preprocessing. The reconstruction algorithm of binocular vision 3D scene includes image preprocessing
multi-thread acceleration stereo matching and AD-Census matching. The algorithm is programmed with Verilog HDL and compute unified device architecture(CUDA).Experimental results show that the system calculating a depth map with 640×480 size is achieved at speed of 7 frames per second
and150 depth maps at 1. 8° intervals can be converted into point clouds