Share
الرؤية المجسمة لتخمين المسافة باستخدام SAD مع مرشح كشف الحواف Canny وتقنيات التعويض == Stereo Vision Distance Estimation Using SAD And Canny Edge Detector With Interpolation Methods
Author name:
زيد خضر حسين
Supervisor name:
رعد حمدان ظاهر
General topic:
Electrical, Electronic and Communications Engineering
Specific topic:
Electronics and Communications Engineering
Degree:
Master
University:
Mustansiriyah University - College Of Engineering - Department Of Electrical Engineering
Language:
English
University location:
Baghdad
First pages:
34T526 - p.pdf
Abstract:
Stereo vision system is a technique for finding depth data for digital images.The stereo vision system is used to obtain a 3D data from 2D scene taken by two optical cameras (left and right cameras); the taken images can be used to find the distance of the objects. A number of algorithms for stereo vision system have been developed and the matching algorithms. This work focuses on the traditional algorithms used the Normalized Cross Correlation (NCC), Sum of Absolute Differences (SAD) and the modified version of SAD algorithm. This method is called the Canny Block Matching Algorithm (CBMA) which can find the disparity map. The proposed algorithm contains two parts; the Canny edge detector and Block matching technique with SAD to determine the disparity map and reduce the execution time, the execution time of the CBMA is between (0.35 - 0.43secs) and neglected the effect of the windows size results in the CBMA algorithm. The interpolation method is used which consists of median filter and interpolation techniques (i.e., the most common techniques : bilinear, 1st order and 2nd order polynomial) to enhance the output results images.The error percentage has been reduced about 2% for the disparity map of the CBMA algorithm after used the bilinear interpolation method with block size [3x3] and the execution time is reduced by the step size windows. MATLAB program has been adopted to write the proposed algorithms codes. While, the interpolation techniques has been implemented using in Microsoft Visual Basic (6.0). The camera calibration and image rectification is used to find camera parameters and to simplify the correspondences search. The system was implemented using two identical cameras with baseline (16cm) to detect the distance objects.
