According to remote sensing image shadow properties,the paper proposes a new shadow detection method based on color space,in order to improve the precision of the shadow detection.In the shadow detection process,firstly the image converts to HSV space because of the high color value,low luminance value and high saturation property.At the same time,define a new quantity M =(S-V)/(H + S + V).Then using spectrum technology and median filter,get the preliminary shadow area.Secondly on the basis of scattering theory about the influence of the blue and the same characteristics of Color,put forward a shadow detection method based on double threshold value.The two threshold values are C3 value in C1C2C3 space and B value in RGB space.To reduce the threshold selection of subjectivity,get the shadow region by means of the unite operation.Lastly extract shadow region combined with removing small area and mathematical morphology processing.In order to proving the effectiveness of this proposed method,do a lot of experiments.Choose a number of the remote sensing images with shadows as the experience images.Because of the limit of the length,3 images are presented.The results show that the proposed method is obviously superior to the traditional histogram threshold value method and Homomorphic filtering method through experiments.At the same time,this new method overcomes the subjectivity for the choice of threshold value,improves the accuracy of the shadow detection.%依据遥感影像阴影的属性,提出一种基于彩色模型的遥感影像阴影检测方法,以提高阴影检测精度.阴影检测过程中,首先将影像转换到HSV空间.根据阴影区域高色调值、低亮度值和高饱和度的特性,定义M=(S-V)/(H+S+V).利用光谱比技术和中值滤波,得到初步的阴影区域.其次依据散射理论对蓝光的影响,以及彩色不变特征,提出结合C1C2C3空间的C3分量和RGB空间的B分量进行双阈值阴影检测.为降低阈值选择的主观性,提出将上述两种方法进行与运算,并结合小区域去除和数学形态学处理,进行阴影提取.最后对多幅带有阴影的遥感影像进行实验.结果表明所提出的方法明显优于传统的直方图阈值法和同态滤波检测法,克服了阈值选择的主观性,提高了阴影检测精度.
展开▼
