Abstract: The traditional image segmentation methods often use color information for mature apple image, but the color difference between immature apple and background is smaller in natural scene, and the effect of segmentation is unsatisfactory. This paper presents an image segmentation based on graph theory with texture information. Firstly, the image is divided into subgraph and calculates the texture information for every subgraph. Secondly, the weighted undirected graph has been constructed with image block as node by using the texture information structured weighted function. Finally, the method of region merging based on graph theory has been used and the results are mergered by absorption secondary simultaneously to finish the segmentation. The experimental results show that this method is better than traditional method and can improve the case of over-detailed division.

Key words: image segmentation, apple image, texture analysis, graph theory

摘要： 针对未成熟苹果图像果实同背景色彩差距较小导致难以分割效果的问题，提出一种融合纹理信息与图论的图像分割方法。该方法先将图像划分为图像块并计算每个图像块的纹理信息，以图像块为节点利用纹理信息构造权函数得到无向带权图，用基于区域合并的图论图像方法进行分割并二次吸收合并。实验结果表明，在对象和背景色彩接近时，该方法比传统分割法效果好，且改善了分割过细的情况。

关键词: 图像分割, 苹果图像, 纹理分析, 图论