中文摘要：

为了提高图像匹配算法的实时性和复杂场号跟踪的稳定性，改进一种基于图像纹理的模板匹配算法。在模板与目标图像中分别提取水平、垂直方向上的二值化纹理特征矩阵，然后根据字节同或逻辑运算分别计算两个方向上的最佳匹配值，最后，通过加权求和得到最终的目标相关置信度。其中逻辑运算可以并行处理，所以计算量将大大减少，具有很强的实时性。又因为跟踪过程中，目标图像必然存在突发性的变化，对模板合理地进行更新是目标跟踪的关键。在上述的图像相似度度量方法的基础上，提出了一种非常有效的模板修正方案，从而使跟踪算法对环境的适应能力和稳定性大大提高。实验结果证明了该方法的优越性。

英文摘要：

To improve the real-time performance of image registration and the stability of target tracking under complex scene, a template-matching method based on image region texture （TIM） was ameliorated. First, binary texture matrixes of horizontal and vertical direction were constructed in the template and target image, and then the prime matching result of the two directions was calculated according to the Alogical XOR logical operation in bytes and the final target correlation credibility was obtained by proportion sum at last. This method have great real-time characteristic and the mount of operation can be reduced greatly because of the logical operation can be done in collateral ways. And in the process of tracking, there must be many sudden changes in the object image ,then the reasonable updating of the reference image will be the key to the tracking problem. In that case, another suitable template updating strategy was presented on the basis of the similarity measurement so the stability and the adaption to the environment has been improved greatly. The experiments show that the method is very promising.