在分析半色调图像特性及传统客观评价方法不足的基础上，提出了分区思想，即将图像分成若干小区域，并对峰值信噪比、归一化均方误差、结构相似度方法进行了改进，考虑了人眼视觉特性，弥补了传统评价方法的不足。研究发现，改进后的方法可以很好地评价图像的质量优劣，提升了主客观一致性。

The characteristics of halftone images and the shortages of traditional objective evaluation methods were analyzed. Partition thinking was proposed, which is dividing image into several small areas and the peak signal to noise ratio, normalized mean square error and structural similarity method is improved. Human visual characteristics were considered to make up the deficiencies of traditional evaluation methods. Study result showed that the improved method can evaluate image quality well and enhance the subjective and objective consistency.