基于颜色直方图和空间信息融合的图像分割算法

A Segmentation Algorithm of Color Image Based on Hue Histogram and Spatial Information

论文提出了一种基于颜色直方图和空间信息相结合的彩色图像分割算法。首先将图像转化为HSI模式,利用高斯滤波器对色度信息进行平滑后获得了合适的色度直方图。基于色度直方图对图像进行了分割,其次,将像素相似性和空间连通性相结合形成了初始分割区域。在此基础上,定义了区域距离函数,实现了图像区域合并操作。再者,从图像合并过程中信息损失和应保留的区域数量之间的关系出发,给出了合并终止规则。最后,针对所提算法进行了仿真研究。

A novel segmentation algorithm of color image based on the combination of hue histogram and spatial information is proposed in this paper.Firstly,the image is converted from RGB model to HSI one,and then Gaussian filtering is applied to the hue histogram to obtain a smoothed hue histogram.Based on the smoothed hue histogram,the image is segmented primarily,and then,according to the combination of the similarity and spatial adjacent connection between pixels,form ing a segmented image.Secondly,a region distance function is defined,by which the growing operation is performed on the primary segmented regions.Thirdly,from the viewpoint of the relation between the reserved regions and the lost information in region merging,the rule of the stop merging is given.Finally,simulations are operated on the segmentation algorithm presented in this paper.