基于分数阶权系数矩阵的静态图像分割算法  

Segmentation algorithm of static image based on fractional power coefficient matrix

作　　者：李忠海 金海洋 邢晓红 陈灿灿 LI Zhong-hai;JIN Hai-yang;XING Xiao-hong;CHEN Can-can(School of Automation, Shenyang Aerospace University, Shenyang 110136, China)

机构地区：[1]沈阳航空航天大学自动化学院,辽宁沈阳110136

出　　处：《计算机工程与设计》2018年第5期1405-1409,共5页Computer Engineering and Design

基　　金：航空科学基金项目(XY201434-2)

摘　　要：边缘停止函数的结构不同导致基于轮廓迭代的静态图像分割算法存在控制边缘精度的能力低、大误差轮廓演变为精确轮廓的速度慢等问题,为此,从改进停止函数的结构入手,以整数阶Sobel算子为例,引入整数阶Sobel算子,通过建立分数阶权系数矩阵定义新的分数阶Sobel算子,提出基于分数阶权系数矩阵的边缘停止函数。改进后的停止函数兼顾分数阶微分图像弱边缘连续性强和整数阶微分能够减少轮廓迭代次数的优势,避免传统停止函数存在的问题。实验结果表明,改进后的停止函数结构能够使弱边缘的识别能力大大提高,大误差轮廓演变为精确轮廓的速度更快,迭代速度提高30%以上。Due to different structures of edge stopping function,the static image segmentation algorithm based on contour iteration has low ability to control edge precision,and low evolving speed of the large error contour to accurate contour and so on.Aiming at the existing problems,starting from the improvement of the stopping function of the structure,the integer order Sobel operator was taken as an example,the integer order Sobel operator was introduced,through the establishment of fractional order weighted coefficient matrix,fractional order Sobel operator was defined,and edge stopping function based on fractional weight coefficient matrix was proposed.The improved stopping function took the advantages of the weak edge continuity of the fractional differential image and the integer order differential into account,reducing the number of iterations of the contour.It avoided the existing problems.Experimental results show that the improved stop function structure can greatly improve the recognition ability of the weak edge.The large error contour evolves to the precise contour faster and the iteration speed is increased by more than 30% .

关 键 词：SOBEL算子 分数阶微分 权系数 边缘停止函数 图像分割 

分 类 号：TN911.73[电子电信—通信与信息系统]