Study on an Improved Robust Algorithm for Feature Point Matching

Review on Corner Detection Algorithms

“…Since the application needs to be detected very quickly, the paper ORB algorithm is used as the corner detection algorithm in this paper. Guo Yongfang et al in 2012 [9], they use Harris corner detection algorithm and Trajkovic fast detection algorithm for feature points matching, and achieved good results. Zhu Hongbo et al in 2011 [10], the authors improve the SIFT algorithm.…”

A New Stable and Accurate Algorithm of Concrete Crack Image Mosaic

“…Many papers have been give the study of matching based on different constraints, such as, the Delaunay triangulation and affine invariant geometric constraint [21], the template matching with arctangent Hausdorff distance measure [22], the coloring based approach for matching [23], the fast and scalable approximate spectral graph matching [24], the optimized hierarchical block matching [25], and improved robust algorithm for feature point matching [26] etc., In this paper, considering of the time-consuming and stability of the total algorithm, we choose the edge cover condition as the constraint of the matching result. On this condition, the edge covered points in one image should cover the edge exactly on the matched image edge.…”

“…Fortunately, the pre-matched homograph matrix M  has provided us the information of the roughly searching region in the matched image. All the above mentioned [21][22][23][24][25][26] pre-matching algorithm are using the self-character of the image contents to find out the best match condition. Although, these pre-defined characters may improve the matching efficient and stability of the matching procedure, for the accurate matching in the crack detecting condition, we need to compare the characters of the each control pixels of the overlapped region.…”

Feature Point Matching with Matching Distribution

“…From the literatures, the techniques to tackle feature point matching can be classified into three groups [10]. The first category is to use only location of the points to find the best correspondences.…”

“…Feature point matching is a very important process for applications that need to locate objects in images or databases such as robot navigation, tele-surgery, and image retrieval [1,2,3]. The feature points are derived from the images, called the reference and input image, of the same scene which may have been acquired at different times, from different viewpoints, and by different sensors.…”