论文部分内容阅读
基于提升方法的整数小波变换的诸多优点,以Harr整数小波变换为例,提出了动态目标跟踪算法。对标准图像进行整数Harr小波变换,并将提升项取整,对提升项的参数用一定数量训练图像进行学习。选取在训练参数平方和为最小意义上的、使整数小波变换后的图像高频部分具有较大值的点作为特征点。对包含有目标的参考图像进行整数小波变换,选择高频分量具有较大值的点,利用训练过的提升项参数使目标和基准图像配准。由于算法采用整数小波变换,使提取具有较好的鲁棒性,从而实现目标发生旋转、平移及尺度变化等的跟踪过程。仿真实验表明,该方法能对动态运动目标进行跟踪。
Based on the many advantages of lifting wavelet integer transform method, taking Harr integer wavelet transform as an example, a dynamic target tracking algorithm is proposed. The standard image is subjected to integer Harr wavelet transform, and the lifting term is rounded, and the parameters of the lifting term are learned with a certain amount of training images. Select the square in the training parameters for the smallest sense, so that the integer wavelet transform the high frequency part of the image has a larger value as a feature point. An integer wavelet transform is performed on the target reference image, a point having a high value of the high frequency component is selected, and the target and the reference image are registered using the trained lift parameter. Because the algorithm uses integer wavelet transform, the extraction has good robustness, so as to achieve the target rotation, translation and scale changes tracking process. Simulation results show that this method can track the dynamic moving target.