Template Matching Method Based on Visual Feature Constraint and Structure Constraint

Abstract

SUMMARYTemplate matching for image sequences captured with a moving camera is very important for several applications such as Robot Vision, SLAM, ITS, and video surveillance systems. However, it is difficult to realize accurate template matching using only visual feature information such as HSV histograms, edge histograms, HOG histograms, and SIFT features, because it is affected by several phenomena such as illumination change, viewpoint change, size change, and noise. In order to realize robust tracking, structure information such as the relative position of each part of the object should be considered. In this paper, we propose a method that considers both visual feature information and structure information. Experiments show that the proposed method realizes robust tracking and determine the relationships between object parts in the scenes and those in the template. key words: template matching, visual feature constraint, structure constraint, linear assignment, genetic algorithm

Robust Moving Object Extraction and Tracking Method Based on Matching Position Constraints

“…We define that matching is successful when all blocks of the template match to the target. The proposed method and the methods in [18] and [8] achieved the highest accuracy.…”

“…By using the Kalman Filter, the number of successful frames is improved for three data: David Outdoor, Sylvester, and Road Bridge. As compared with the method in [8], the proposed method realized more robust template matching. Examples of the template matching results are shown in Fig.…”

“…The matching position constraints are applied as for object tracking. In Table 1, we compared the proposed method with six existing methods in [14]- [18], and [8] for David Indoor [19]. We define that matching is successful when all blocks of the template match to the target.…”

“…Moreover, in Table 2, we show a comparison with the method in [8] for six data including David Indoor. In these experiments, we also adopted Kalman Filter [13] to predict the object location.…”

“…However, this problem cannot be solved strictly because we have to solve the quadratic assignment problem. Therefore, it is approximated by using the steepest descent method [7], a genetic algorithm (GA) [8], successive convexification [9], and the iteration of visual feature optimization and shape optimization [5].…”