Research On Binocular Vision Technology For Fully Autonomous Intelligent Robot

The information of fully autonomous intelligent robots is processed by thechip-on-board completely, so most of this are designed by embedded system. Thereis a wide gap on processing speed and storage capacity for fully autonomousintelligent robots which is designed on embedded system relative to ordinary PC, solet the image processing algorithms run better in embedded systems is a bigchallenge.Currently, almost all of the fully automous intelligent robots apperceive thesurrounding environment by using a single camera. Though the monocular visiondoes have its advantages relative to the binocular stereo vision, and it existssignificant deficiencies also. This thesis contrast the monocular and binocular visualperformance from the region of interest segmentation, distance accuracy, timecomplexity firstly and summary the following conclusion. Monocular vision hassmall amount of information and only deals with one image each time, so it has fastarithmetic speed and often be used in the simple scene. But the3D information ofthe actual scene is lost in the process of projection. At the same time the monocularvision system stringent requirements on surrounding environment, such as light,color, saturation. Some unendurable judgement mistakes will appear if the scenebecomes a bit more complicated. Relatively, binocular stereo vision overcomesthese problems very well, meanwhile it increases a lot of computation, so the realtime autonomous control decision of fully autonomous intelligent robot which isdesigned on embedded sytem becomes a challenge.In this thesis, whith demand of the need to determine the coordinates of theregion of interest for intelligent robot, proposed a efficiency coordinate calculationmethod which based on judge the longest straight-line, this method is robustly andhight accuracy. By in-depth analysis and research the stereo matching through alarge number of experiments, decided to adopt the SAD similarity measure forstereo matching. Proposed and prove the same centroid of image principle, andapply this principle to the proposed efficient interval stereo matching algorithm forintelligent robot, this method can retain most effective information of the image.Design a system suitable for intelligent robot binocular stereo vision which issummed up by researching every key stage’s algorithm and comparing relatedexperiments thoroughly, achieve good result.