摘要
本文研究了一种基于形态学处理和纹理特征合并的水域算法,该方法首先对形态梯度图进行滤波处理,以获得较好的参考图像,然后又利用图像的直方图势函数对图像做标记。最后,为了获得整体的目标,我们对上面所得到的分割结果进行了区域一致性、纹理一致性和对比度的检验,来合并用水域算法获得的分割结果,以获得更好的分割效果,并通过试验验证了算法的有效性。
This paper presents a target segmentation algorithm based on morphological processing and texture analysis. Firstly an opening-closing bank is performed on the gradient image in order to get a better reference image, then a label based on the potential function is made to image. Finally, in order to get an integral target, we use the criterion with respect to the homogeneity and contrast of the textures to merge the regions, and get a better segmentation result.
出处
《计算机工程与科学》
CSCD
2005年第11期28-30,44,共4页
Computer Engineering & Science
关键词
数学形态学
水域分割
直方图势函数
形态学滤波组
mathematical morphology
Watersheds
potential function
opening-closing filter
