图像质量评价综述

Overview of Image Quality Assessment

图像质量评价是图像处理领域内一项很有意义的研究课题。图像质量评价分为主观图像质量评价方法和客观图像质量评价方法。主观图像质量评价方法包括双刺激损伤分级法、双刺激连续质量分级法、单刺激连续质量分级法。相比主观图像质量评价方法来说,客观评价方法应用更准确、广泛。客观图像质量评价方法可分为全参考评价方法、半参考评价方法和无参考评价方法,目前全参考评价方法较为成熟,众多方法中最常用的是均方误差(MSE)法和峰值信噪比法(PSNR)。而半参考和无参考评价方法则处于初级阶段。通过对主观、客观图像质量评价不同方法的研究、分析、对比得出相应方法的优缺点。对图像质量评价方法进行展望。

Image qual ity evaluation is a very interesting research topic in the field of image processing. Image quality evaluation is divided into subjective image quality evaluation method and objective image quality evaluation method. Subjective image quality evaluation methods include double stimulation damage grading method, double stimulation continuous mass grading method, single stimulation continuous mass grading method. Compared with the subjective image quality evaluation method, the objective evaluation method is applied more accurately and extensively. The objective image quality evaluation method can be divided into full reference evaluation method, semi-reference evaluation method and non-reference evaluation method. At present, the whole reference evaluation method is more mature, and the most commonly used method is the mean square error （MSE） method and the peak signal to noise ratio Law （ PSNR）. While the semi-reference and non - reference evaluation methods are in the in itial stage. Through the research, analysis and comparison of the different methods of subjective and objective image q u al ity eva lu ation, the advantages and disadvantages of the corresponding methods are obtained. The method of image quality evaluation is forecasted.