摘要
在合成孔径雷达(Synthetic Aperture Radar)自动目标识别中,图像分割的好坏直接影响目标的识别性能。本文在最大模糊熵分割方法的基础上,根据图像目标和背景内部像素灰度值的一致性和集中性,提出了一种新的图像分割隶属度函数,从而得到最小模糊熵分割方法,然后将最小模糊熵作为遗传算法的适应度函数应用于SAR图像,进行全局快速的最优阈值寻找。实验结果表明,由于最小模糊熵的抗噪能力强,将其作为遗传算法的适应度函数后,能够更有效地克服SAR图像中的乘性噪声,分割后的噪声点明显减少,图像目标清晰,分割效果明显优于最大模糊熵分割方法。
In the process of SAR automatic target recognition, image segmentation plays a key role. Based on the Maximum Fuzzy Entropy image segmentation, a new function of image segmentation is presented in this paper according to the consistency and concentricity of the pixel' s gray levels inside the image objects and the background, then the minimum fuzzy entropy image segmentation is achieved. Taking the minimum fuzzy entropy as the adaptability function of genetic algorthm, the optimum threshold value in SAR image is sought. The experimental result shows that genetic algorithm based on the minimum fuzzy entropy image segmentation has better effect than that based on maximum fuzzy entropy image segmentation.
出处
《航空兵器》
2009年第1期30-33,共4页
Aero Weaponry
关键词
合成孔径雷达
模糊熵
图像分割
遗传算法
synthetic aperture radar
fuzzy entropy
image segmentation
genetic algorithm