图象压缩可以看作是在码率和图象质量视觉感知失真之间的一种折衷。图象质量视觉感知失真的正确评价直接关系到图象压缩编码算法的设计与优化。传统的图象质量评价方法不能有效地反映出对人眼对图象失真的视觉感知。由于人眼是最终的信息接收者，我们在人眼视觉模型（HVS）的基础上建立了通用的评价模型，并采用了B样条小波、误差分割等技术，来反映人眼在二维随机域上幅度非线性衰减、二维视觉敏感度带通和视觉多尺度多方向性的特点。图象质量评价将向视觉生理心理试验、HVS模型的简化及通用模型的推广三个方向发展。

Image compression could be regarded as a compromise between the bit ratio and perceptual distortion. Image quality evaluation plays a very important role in the design, optimization and evaluation of an image compression algorithm. Classical image quality evaluation methods such as MOS, MSE and PSNR can not represent distortion perceived by human eyes. New measurements are based on HVS (Human Visual System). They are generally defined as a function of Amplitude Nonlinear it y, CSF, Detection M echanisms. We build a general evaluation model on HVS by B spline wavelet and error signal segmentation technology. The future of image quality evaluation will be directed to psychology and physiology experiments, the simplification of HVS and generalization of evaluation model.