Adaptive Dual-threshold Algorithm for Segmentation of Shadows Underneath Vehicles

Min Li, Xing Li, Yujie He, Yuhang Xing


Shadow segmentation is significant in vehicle detection. This study focused on improving the adaptability to illumination changes. An adaptive dual-threshold algorithm was proposed for the segmentation of shadows underneath vehicles. The upper boundary of the segmentation threshold was set based on histogram distribution analysis. One threshold was calculated using the Otsu method in the range between the lowest grayscale and the upper boundary for the gray image. Another threshold was set based on the statistical analysis of Sobel edge-enhanced images. This dual-threshold algorithm was utilized for comprehensive shadow segmentation. Experimental results show that the proposed method can effectively set the threshold and meet the requirements of shadow segmentation in different illumination circumstances with improved adaptability.

Full Text:



  • There are currently no refbacks.