A Matching Algorithm Based on Multi-Resolution and Edge Enhancement Using the Optimum Size of Image

摘要

This paper analyzes the relation between the size of real time image and the matching time, and a optimum size of image is improved while maintaining a high probability of correct matching. An accelerated normalized product correlation similarity measure is proposed for practical application. In order to enhance the edge feature of image and increase the probability of correct matching, this algorithm enhances the edge of image using the sobel operator. To use image matching in cruise missile, an approach which can accelerate the normalized product correlation matching algorithm is designed and implemented by adopting "from coarse to precise" control strategy. When in the coarse phase, 5 matching-point are choosed, which can contain the primary point by the simulation experimental results. When in the precise phase, a proper switching-level Q is proposed, and the searching is performed around each matching-point. If correlation coefficient is above Q, the searching process ends. The simulation experimental results demonstrate the efficiency and practicability of the matching control strategy and the excellent adaptability of the proposed algorithm to grayscale distortion, noise disturbance, geometric rotation. The improved algorithm maintains a high probability of correct matching, and the running time reduces 1/5 to 1/8, or even more.