摘要
遗传算法在复杂大空间搜索近似最优值有着很好的效果,利用遗传算法的优势,应用其解决图像分割问题。图像分割中,区域间差别度和区域内相似度是评价图像分割的重要因素,用遗传算法优化图像区域间差别和区域内相对相似度,获得高质量的图像分割结果。为了提高算法效率,采用贪心方法进行图像预处理,以及最小生成树初始化来减小算法的规模和搜索空间。实验证明采用遗传算法在图像分割问题可取得有效的结果。
Generic algorithm is quite effective while searching near-optimal solutions in complex and large spaces.In this paper we applied generic algorithm in solving the image segmentation issue with its advantages.The inter-regional differences and the inner-regional similarities are two important factors reflecting the effect of image segmentation.Our method achieves image segmentation effect with high quality through simultaneously optimising the inter-regional differences and the inner-regional relative similarities of the image with generic algorithm.Besides,we use greedy algorithm to pre-process the image,and the minimum spanning tree are also used to decrease the computing scale and search space of the algorithm.Experiments proved the effectiveness of the generic algorithm in solving image segmentation.
出处
《计算机应用与软件》
CSCD
2011年第3期237-239,256,共4页
Computer Applications and Software
关键词
彩色图像分割
遗传算法
图像预处理
最小生成树
Colour image segmentation Genetic algorithm Image pre-processing Minimum spanning tree
