基于支持向量机的无参考遥感图像质量评价方法

针对遥感图像主观评价方法的低效率以及常用客观评价方法无法充分考虑人眼对图像的感知特性的问题，文章提出了一种基于支持向量机的无参考遥感图像质量（quality）评价方法。首先建立遥感图像主观评价库，然后在不需要图像失真信息的基础上，利用支持向量机（SVM）将图像的失真类型分为三类，并对每类进行单项评价，再通过加权得到遥感图像的总评分，最后将本文方法、信噪比与信息熵的评价结果回归到主观评价空间并进行对比。实验证明，文章所提方法能客观地评价遥感图像的质量，且优于信噪比和信息熵两种质量评价方法，其结果与人眼视觉感受相符。

Assessment Method for No-reference Remote Sensing Image Quality Based on Support Vector Machine

In view of low efficiency of subjective assessment method for remote sensing image and lack of full consideration of human eye perception of common objective assessment method for the image features, this paper proposes a no-reference remote sensing image quality assessment method based on support vector machine. Firstly, a subjective assessment library for remote sensing image is established, and then,without image distortion information, we use support vector machine to classify image distortion into three categories, and compute individual evaluation for each category. The final remote sensing image quality is obtained by probability-weighted summation. Finally, the results of the method proposed by the paper, the signal-to-noise ratio and the information entropy are regressed back to the subjective assessment space and compared. In the paper, the proposed method is demonstrated a good objective assessment method of remote sensing image quality, and this method is superior to the assessment methods of signal-to-noise ratio and the information entropy, whose consistent with human visual experience.