多信息融合的视频运动目标阴影去除算法

Multiple information fusion of video moving target shadow removal algorithm

针对传统HSV空间阴影去除模型中阈值难以确定、计算复杂及检测效率较低等问题,在对传统运动目标阴影去除算法进行深入研究的基础上,首先融入一阶梯度信息对传统HSV空间阴影去除模型的不足之处进行针对性改进,然后在此基础上融入反射比不变量提出了一种多信息融合的视频运动目标阴影去除算法。该算法在改进HSV空间阴影去除算法的基础上,进一步引入阴影候选像素及其对应背景区域像素的反射比不变特性来实现阴影区域更为精确的检测,从而有效区分并去除运动目标的阴影像素。实验结果表明,该算法在实际应用中具有较高的有效性和通用性。

In view of the series of problems that it is difficult to determine the threshold value in the model of removing the traditional HSV space shadow, and complex to accomplish the computation as well as the improvement of low detection efficiency, which is based on the in-depth algorithmic study for the traditional shadow removing of a moving target, focuses on the purposeful improvement of the first-order gradient information deficiencies in the model of removing the traditional HSV space shadow. There- after, it puts forward a new shadow removal algorithm of multiple video information fusions by introducing the reflectance invariants into it. The experiment results have illustrated that this algorithm possesses a higher validity and universality in the practical appli- cations. It can not only improve the algorithm of the traditional HSV space shadow removal, but also realize the more exact detec- tions in the shadow area through the integration of reflectance invariants characteristics between the shadow candidate pixels and the background pixels. In the end, this algorithm is able to effectively distinguish and remove the shadow pixels of the moving object.