改进模板匹配的快速识别算法

A fast recognition algorithm for improved template matching

为解决噪声导致的模板图像误匹配问题以及提高算法的识别速度和准确率,提出一种改进模板匹配的快速识别算法。该算法使用改进的低通滤波器对图像进行降噪处理,并使用改进的加权Hamming Distance,对图像进行相似度检测,降低算法的复杂程度,最后使用阈值和同伦骨架加快算法识别的运行速度。通过对不同背景下的噪声图像进行实验,该算法在目标识别过程中有很好的识别效果,对噪声有很好的鲁棒性,并且能够改善算法的运行效率,从而证明了该算法在目标识别过程中的有效性。

In order to solve the problem of mismatch caused by noise,and to improve the speed and accuracy of recognition algorithm,a fast recognition algorithm with improved template matching was proposed.A novel low-pass filter was used to perform noise reduction on the image.Then,an improved weighted Hamming Distance was used to perform similarity detection on the image to reduce the complexity of the algorithm.Finally,the threshold and homotopy skeleton algorithms were used to speed up the recognition speed.By experimenting with noise images in different backgrounds,the proposed algorithm had a good matching effect in the target recognition process,and had good robustness to noise,and improved the efficiency of the algorithm.Thus the effectiveness of the algorithm was proved.