基于HASH感知和小波变换的目标跟踪算法

为了解决视频跟踪过程中,由于运动目标在实际运动中常存在旋转、缩放等形变导致的目标丢失问题,提出了一种融合了图像HASH值、小波变换和模板匹配算法的目标跟踪算法。该方法利用背景差分法得到运动目标模板,通过对图像信息进行小波变换得到低频子图像,并在低频子图像中进行全局搜索，确定最佳匹配区域。模板更新策略利用图像间HASH距离,判断图像间的相似性决定是否更新模板。实验仿真结果证明,该算法在运动目标存在缩放、旋转等形变时,仍可以准确跟踪目标。

Target tracking algorithm based on perceptual HASH and wavelet transform

To solve the targets missing caused by the rotation scaling or some other deformation of the targets which are common in the actual movement during the video tracking process, a tracking algorithm combing the image HASH, wavelet transform and template matching algorithm is proposed. The template using in this method is extracted by the background difference method. Through the wavelet transform of the image information, low frequency sub-images can be obtained, and the best matching region is determined according the global search. Using the HASH distance between images, the template updating strategy to judge whether to update the template or not is concluded. The simulation results prove that this algorithm can track the target accurately when the moving target is covered or has deformation of scaling and rotation.