基于归一化RGB色彩模型的阴影处理方法

Shadow processing method based on normalized RGB color model

通过分析影像上阴影区域的属性,提出了一种基于阴影属性的高分辨率遥感影像阴影检测和去除算法。利用阴影区域蓝色分量偏高的特性,对归一化B分量和原始B分量进行阈值检测,并结合小区域去除和数学形态学处理,得到较精确的阴影区域;然后,分别在RGB空间和HSI空间对各个独立的阴影区域与其邻近的非阴影区域进行匹配,完成阴影去除操作;最后沿着阴影边界做一次中值滤波以减轻边缘效应。仿真结果验证了算法的有效性,并且显示在HSI空间获得了更好的补偿效果。

A new property-based shadow detection and shadow removal algorithm is proposed for high resolution remote sensing image after analyzing the properties of shadow regions. Threshold detection is performed on normalized blue band, and original blue band based on shadow regions has higher pixel value than non-shadowed regions in blue band. Combined the two threshold detection results, an original shadow mask is obtained. Then small region removal and morphological algorithm are used to get accurate shadow mask. Each shadow region is matched to its adjacent non-shadowed region on RGB （Red, Green, Blue） color space and HIS （Hue, Intensity, Saturation） color space respectively to finish shadow removal. At last, median filtering is performed along the edge of each shadow mask in order to reduce the effect of noise. The experimental results demonstrate that the algoritbm performs well and a better compensation result is obtained in HSI color space.