一种基于先验知识的分水线彩色图像分割方法

Watershed Color Image Segmentation Based on Prior Knowledge

提出了一个结合彩色信息的分水线图像分割方法,在LUV颜色空间的分水线图像上,使用贝叶斯推理的图像分割方法。对图像进行LUV彩色空间上的分水线变换,在分水线图像上进行基于彩色信息的各个区域能量的计算,通过选择最小能量的目标,依次找出最理想的目标区域。设计了一个先验密度惩罚图像当中分水线变换后的相似的区域,图像分割进而变成对目标子集的最大后验估计。逐步地找出最理想目标区域和背景区域。这一方法同时结合使用了彩色信息和空间信息;可以有效地解决分水线变换后的过分割问题,利用了先验知识和彩色信息。实验结果显示,该方法有较好的分割结果。

This paper proposes a new image segmentation algorithm based on watershed transformation combined color information and uses Bayesian inference on this watershed image. It transforms RGB color image to LUV space, uses watershed transformation based this color image. This paper calculates the energy of the label image result from the color image watershed transformation by designing a prior density that penalizes the area of homogeneous parts in images. The segmentation problem is the maximizing a posteriori estimation of the set of object area result from the watershed labeled, Then could find the opllmal area of object, and the olher area of the image looked upon background area. This algorithm not only solves the over-segmentation problems of watershed transformation, but also uses color information and prior knowledge. The experiments indicate the algorithm is effective for image segmentation