Automatic Image Mosaic Technique Based on Improved RANSAC Algorithm
Shiliang Steven Zhang · Journal of Sanming University · 2012
The key of image mosaicing technique lies in both image registration and the efficiency of the algorithm.According to the characteristics of the traditional image feature-matching algorithm with the disadvantages of a large quantity of data of corner point pairs and time-consuming calculation,this paper applies K-means clustering algorithm to the image-matching algorithm and proposes the improved method.This method uses K-means clustering algorithm to make the corner points extracted from the image clustered and grouped and improves the efficiency of the traditional algorithm through a pre-detection model and a stochastic grouping method.The experimental results show that,compared with the traditional RANSAC algorithm and under the condition of keeping higher precision and robustness,this algorithm greatly improves the calculation efficiency,whereby contributing to the efficiency of automatic image-mosaic.