A stereo matching algorithm using multi-peak candidate matches and geometric constraints

Gray cross correlation matching technique is adopted to extract candidate matches with gray cross correlation coefficients less than some certain range of maximal correlation coefficient called multi-peak candidate matches. Multi-peak candidates are extracted corresponding to three closest feature points at first. The corresponding multi-peak candidate matches are used to construct the model polygon. Correspondence is determined based on the local geometric relations between the three feature points and the multi-peak candidates. The disparity test and the global consistency checkout are applied to eliminate the remaining ambiguous matches that are not removed by the local geometric relational test. Experimental results show that the proposed algorithm is feasible and accurate.

A stereo matching algorithm based on SIFT feature and homography matrix

Aiming at the low speed of traditional scale-invariant feature transform(SIFT) matching algorithm, an improved matching algorithm is proposed in this paper. Firstly, feature points are detected and the speed of feature points matching is improved by adding epipolar constraint; then according to the matching feature points, the homography matrix is obtained by the least square method; finally, according to the homography matrix, the points in the left image can be mapped into the right image, and if the distance between the mapping point and the matching point in the right image is smaller than the threshold value, the pair of matching points is retained, otherwise discarded. Experimental results show that with the improved matching algorithm, the matching time is reduced by 73.3% and the matching points are entirely correct. In addition, the improved method is robust to rotation and translation.

A Novel Method for Multi-image Matching Synthesizing Image and Object-space Information

A novel method for multi-image matching by synthesizing image and object-space information is proposed. Firstly, four levels of image pyramids are generated according to the rule that the next pyramid level is generated from the previous level using the average gray values of the 3 by 3 pixels, and the first level of pyramid image is generated from the original image. The initial horizontal parallaxes between the reference image and each searching image are calculated at the highest level of the image pyramid. Secondly, corresponding image points are searched in each stereo image pair from the third level of image pyramid, and the matching results in all stereo pairs are integrated in the object space, by which the mismatched image points can be eliminated and more accurate spatial information can be obtained for the subsequent pyramid image matching. The matching method based on correlation coefficient with geometric constraints and global relaxation matching is introduced in the process of image matching. Finally, the feasibility of the method proposed in this paper is verified by the experiments using a set of digital frame aerial images with big overlap. Compared with the traditional image matching method with two images, the accuracy of the digital surface model (DSM) generated using the proposed method shows that the multiimage matching method can eliminate the mismatched points effectively and can improve the matching success rate significantly.

A Robust Algorithm Based on Object Contours and Order Matching for Disparity Map Post-Processing

Based on the feature of stereo images' content and the property of natural objects, we redefine the general order matching constraint with object contour restriction. According to the modified order matching constraint, we propose a robust algorithm for disparity map post processing. Verified by computer simulations using synthetic stereo images with given disparities, our new algorithm proves to be not only efficient in disparity error detection and correction, but also very robust, which can resolve the severe problem in the algorithm proposed in Ref. that if there are large differences among the depths of objects in a scene, the algorithm will make mistakes during the process of disparity error detection and correction.