新型遥感立体图像编码质量评价方法

提出一种新型遥感立体图像编码质量客观评价方法．全面考虑造成立体图像失真的4个主要因素:由辐射特性畸变、纹理畸变、相关性损失和几何特性畸变，建立了立体图像编码质量客观评价数学模型．实验证明，该方法性能高于常用的均方误差和峰值信噪比图像质量评价方法，用此法对遥感立体解压缩图像的质量进行评价，其结果与人眼的感知评价保持较好的一致，该方法与主观的平均评价分数法具有很强的相关性．

New method for coding quality assessment of the stereo remotely sensed imagery

The method for coding quality assessment of the remotely sensed imagery plays an important role in studying the compression algorithm of the remotely sensed imagery. A new coding quality assessment objective method for the stereo remotely sensed imagery, which is based on the characteristics of the images and the practical application in the stereo mapping, is presented in this paper. The mathematical model of coding quality objective assessment of the stereo image is established by taking full account of the stereo image distortion caused by the four main factors: the radiometric character distortion, texture distortion, loss of correlation and geometrical character distortion. Experimental results show that this method is superior to the common methods such as MSE(Mean Square Error) and PSNR(Peak Signal to Noise Ratio). By using this method, a satisfactory result has been obtained, which is consistent with that from the human visual perception. And the method has strong correlation with the subjective method MOS(Mean Opinion Score).