Three-dimensional Shape Measurement Using Improved Binary Spatio-temporal Encoded Illumination and Voting Algorithm

Some regions of objects will be measured incorrectly or can't be measured in optical 3D measurement system based on coded structured light, due to occlusion, shadow, transfer function of measurement system and noise etc. To obtain 3D data as much as possible and as correctly as possible, we proposed a method using improved binary spatio-temporal encoded illumination and voting algorithm. Firstly, the BSE pattern is improved with a redundancy encoding method. One code is assigned to two adjacent sections and distinguished with their temporal coordinates. The redundancy encoding method provides more cues for code correcting and retrieving. Secondly, symbols are estimated according to four coding cues - code redundancy, continuity of stripes, intensity variation in temporal direction and neighbor symbols in sequence. Finally, a voting algorithm is adopted to obtain final symbols. A plaster model of human head was measured to validate the method. The experimental results reveal that more valid points can be obtained and the reliability of the decoding results is improved.