基于形态学与灰度模态分析的阴影去除方法

针对采集图像中物体附近可能会形成阴影而造成干扰的问题,提出一种利用形态学修复方法和经验模态分解（empirical mode decomposition,EMD）技术实现阴影去除与目标图像的准确提取的方法。采用背景差分和二值图像形态学修复方法检测出含有目标及其阴影的区域,利用经验模态分解方法对该检测区域的灰度直方图曲线进行处理,获取灰度模式变化信息,将搜寻到关于目标及其阴影分割的双阈值水平,结合到形态学修复以实现阴影区域的去除,并以汽车及行人图像处理的实验进行验证。实验结果证明：图像中的目标阴影得到有效去除,该方法具有良好的适应性,能够获得准确的目标检测效果。

Shadow Removal Method Based on Morphology and Grayscale Mode Analysis

A fusion analysis method was explored for the shadow interference problem near the objects in the image.The mathematical morphology and empirical mode decomposition（EMD） were adopted to achieve the shadow removal and accurate extraction of the target image. The background subtraction and mathematical morphology were applied for the binary image, and to detect the region which contained the target and its shadow. Then, the empirical mode decomposition method was utilized to acquire the grayscale mode information from the gray histogram curve of detection region. Hereafter,the dual-threshold level could be found to distinguish the target and its shadow, the morphological technology was combined to repair the binary image. Finally, the shadow was removed. The images of car and pedestrian was tested to prove the result. The experimental results showed that the shadow of target in images could be removed effectively. The proposed method is well adaptable, and it can achieve the accurate target detection purpose.