基于超像素分割的快速移动阴影检测方法

Moving shadow detection method based on super pixel segmentation

针对快速移动车辆阴影检测问题,提出了一种基于超像素分割的移动车辆阴影检测方法。首先基于简单线性迭代聚类方法对原始图像进行同类质超像素分割处理,并将分割单元作为检测集合,在HSV色度空间内完成移动阴影的初步检测;接着,基于局部二值模式(LBP)纹理不变性消除误检像素,并采用区域生长方法实现边缘像素的修正,保证移动阴影干扰车辆的检测正确性和完整性。实验结果表明,在高速公路快速移动车辆的阴影检测中,具有较高的检测精度。

Aiming at the shadow detection of the fast-moving vehicle,a new method for moving vehicle shadow detection based on super pixel segmentation is presented in this paper. Firstly,the original image is divided based on simple linear iterative clustering method,and the segments result is used as the test set. The preliminary shadow detection is finished based on the HSV color space. Second,the mistakenly identified pixels are estimated by using LBP( local binary pattern) invariance texture,and the edge pixels correction is realized based on the region growing method. This method ensures the moving shadow interference detection correctness and completeness of the vehicle. The experimental results show that the proposed method has higher detection accuracy in the shadow detection of fast moving vehicles.