曲线演化的图像分割算法改进研究  

A Research on Improving Image Segmentation Algorithm Based on Curve Evolution

作　　者：曾金发[1] 曹功坤[1] 

机构地区：[1]江西财经职业学院,江西九江332000

出　　处：《计算机仿真》2011年第3期292-296,398,共6页Computer Simulation

摘　　要：研究提高曲线演化的图像自动检测效果问题。由于图像识别度低、准确度低,传统Chan-vese活动轮廓模型(C-V模型)不能检测到远离活动轮廓线且与平均灰度值相差大的边缘。图像分割算法采用水平集演化曲线外图像的目标和背景的灰度加权平均值,通过调节权重值,使演化曲线能准确快速收敛于远离平均灰度强度的图像边缘上。该算法具备拓扑变化能力,分割速度快,能克服原C-V模型不能检测到边缘缺陷,加速图像分割的收敛速度,提高分割效果。This essay has made a research on improving the auto-detection effect for images based on curve evolution.Since the image recognition and accuracy are low,the traditional Chan-vese active contour model（C-V model） can not detect the edge far away from the active contour lines and different from the average gray value.By using the gray weighted average of object and background out of image of evolution curves of level collection and adjusting the weights,the image segmentation algorithm can converge the evolution curve accurately and quickly to the image edge far away from the average gray intensity.The algorithm has the ability of topology changes.It can split fast,overcome the defects that C-V model can not detect the edge,accelerate the convergence speed of image segmentation and improve the segmentation effect.

关 键 词：边缘检测 活动轮廓 水平集 曲线演化 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]