三种典型模板匹配算法性能比较

为了在图像识别、配准和拼接等具体问题中能够针对不同条件来合理选取适当的匹配算法,详细介绍三种典型模板匹配算法,即基于索引表的图像匹配算法、基于图像相关性的图像匹配算法和序贯相似性检测算法(Sequential Similarity Detection Algorithm,SSDA),并从实时性、鲁棒性和精确性等角度出发,对其匹配性能进行差异分析。采用分组实验的方法对匹配算法的各项性能指标进行测试和比较,结果表明,三种算法都具有良好的精确性,其中基于索引表的算法和序贯相似性检测算法具有较好的实时性,而基于图像相关性的算法在噪声环境下具有良好的鲁棒性。

Comparison and research on three typical template matching algorithms

In order to reasonably select an appropriate algorithm for different conditions in particular problems such as image recognition,registration and mosaicking,three typical template matching algorithms,i.e.index table algorithm;an algorithm based on visual correspondence;and sequential similarity detection algorithm （SSDA ）,are explained in details in this paper. Differences analysis of matching performance of these algorithms is processed on their robustness,speed and precision.Experimental simulations data under different circumstances show that all three algorithms perform perfectly in precision,the index table algorithm and SSDA perform better in speed and the algorithm based on visual correspondence performs well on robustness in noise circumstance.