基于结构相似度的SAR干扰效果评估方法

摘要

传统的合成孔径雷达(SAR)图像质量评价指标没有考虑像素点间的相关性和人眼视觉系统感知特性,导致主客观不一致的现象时有发生.提出了一种主客观相结合评估方法,该方法不仅巧妙地利用了结构相似度( SSIM)作为主客观评价的连接纽带,而且结合局部方差相似度与图像高频信息相对应的特点,通过非线性合并得到了结构相似度的修正模型.仿真结果表明,该方法较之传统评估手段具有明显的优势,不仅能够更好地反映出干扰对图像细节的影响,而且能够与人类视觉系统保持高度一致,进一步提升了整套评估系统的效率和准确率.%Without considering correlation of pixels and the perceptual characteristics of human vision system (HVS) , traditional synthetic aperture radar (SAR) image quality evaluation indexes are limited , so that the subjective and the objective means can not come to consistent. A new evaluation method which combines the characters of the subjective and the objective is proposed. This method can not only make good use of structure similarity as juncture between the subjective and the objective evaluation methods, but also utilize the character of local variance similarity which matches with high frequency information in image. Finally, the modified evaluation model is obtained by nonlinear combination. Simulation results show that the new method has an obvious advantage over traditional evaluation methods. Besides reflecting the influences of jamming on ROI in the image, it can keep identical to HVS, and furthermore, it improves the efficiency and accuracy of entire evaluation system.