Stereo vision-based 3D measurement method using segmented phase coding without phase unwrapping

3D measurement method without phase unwrapping has attracted extensive attention due to its requirement for fewer fringe patterns. In a binocular structured light measurement system, the measurement range is determined by the common area of the three perspectives. This results in a high similarity between the deformed fringe patterns captured by the two cameras. Based on this characteristic, this paper proposed a binocular stereo 3D measurement method without phase unwrapping that uses periodic segmented phase coding fringe for auxiliary matching. Firstly, the problem of segmented phase codes matching is regarded as a sequence similarity issue, and the weighted Longest Common Subsequence (LCS) algorithm is used for row-by-row alignment. Subsequently, the monotonic regions of each wrapped phase are matched using the monotonicity, and the partial disparity is obtained from the wrapped phase value. Finally, the disparity in the jump areas is obtained from the π phase shifting wrapped phase. This method requires three sinusoidal phase shifting images and two segmented phase coding images, and there is no need to correct the periodic misalignment. The projector does not require calibration. Experiments have verified the effectiveness of this method, with a root mean square error of 0.052 mm in the measurement of the standard ball.