Research Illumination Field Adaptive Threshold Segmentation And Structured Light Image Decoding Method

In recent decades, computer vision becomes one of hot directions in researchdomain that there is a widely application in intelligence domain which like travelingrobot, industrial control and security system for tracking. As one branch of binocularstereo vision, the technology of structured light recognize, demarcate andthree-dimensional reconstruct by projecting mode symbol on targeting object whichcan be recognized the thee-dimensional object by computer. Because of simplenessand reliability of structured light for coding, so there is more and more practicalapplication for structured light which occupies larger advantage.This thesis is mainly to improve the decoding method which exists somethingnot enough in research group early days. According to the basement of mode symbol,it proposes a method of adaptive threshold segmentation which can overcome theintensity of illumination field and elaborates new decoding method of array of “M”structured light symbol. This method is good to adapt to illumination field, and it canprovide data support for three-dimensional reconstruction in later period. Now, thepassage will unfold for several sections in the following text.1．Research for adaptive threshold segmentation. This passage adopts themethod which can solve question of threshold segmentation to avoid repeatedthreshold segmentation. The algorithm take threshold segmentation by environment ofdistributional mode symbols in valid unit area that it deals with the handling ofnormal estimate of points and the basement of distributional characteristic for modesymbol. Via the whole situation to scan valid range, the algorithm will do thresholdsegmentation efficiently.2．Processing for segment broken. If there is something wrong with one ofmode symbol recognition, it will cause the misrecognition of range region around themode symbol. In this passage, the algorithm makes use of backtracking method toimprove formation of segment and to get rid of nonstandard pixel point information,then to form new run-list again. It enhances independence and robustness of algorithmfor unit region of mode symbol. 3．Unique window-based for mode symbol’s matching. If there is existedmisrecognition which is caused by breakdown or disturbance, it will make use of theunique window-based for mode symbol’s matching to pick up the actual position ofcoordinate where is in symbol template. With result of matching and recognition, thesystem of algorithm send new coordinate information to symbol’s queue which haverecognized instead of no recognition.4．The discussion and study about damaged mode symbol’s recognition. Inactual processing, this passage takes the non-recognition region as extra mark that itcan use pixel to form white or black segment to intersect which is to solve thequestion of part pixel hiatus and symbol damaged in fringe for some photos digitalcamera takes. Finally, computer can estimate damaged pixel region whether isinterference signal or damaged symbol, and it also can get measuring point relevantlyin rational position.