In today’s society, With the rapid development of digital image processing technology, which changes rapidly, the importance of image quality evaluation is more and more outstanding and playing a crucial and irreplaceable role in the field of image processing. At present, a lot of mainstream objective image quality evaluation approaches have been proposed, such as the Peak Signal-to-Noise Ratio(PSNR), the Mean Square Error(MSE) and so on, the objective evaluation criterion is simple and easy to implement but purely statistical data on Mathematics and physics, without the consideration of human mental and visual experience, so it does not fully comply with the visual characteristics of human eyes. International criterion is the Mean Opinion Score(MOS) which called MOS, namely the humanvisual satisfaction of images, but this method is very time-consuming,and need a large number of participants, and the evaluation results are influenced by factors such as the psychological quality of the observer, With the harsh conditions,it is not easy to realize.In view of this, the researchers put forward a lot of new objective quality evaluation criterion combined with human visual characteristics. How to measure the image quality evaluation criterion, should be determined according to the human vision judgment results. In this paper, based on the MOS database, with the least square method to calculate the correlation coefficient to measure the objective evaluation and subjective evaluation to determine the degree of agreement of MOS, the main works are follows:1. Introduces the MOS image database of U-Texas at Austin (America Dezhou University at Austen Laboratory for image and video engineering) and IVC (International Video Center), and introduces the MOS image database of IVC particularly, the MOS image database contains10pieces of original image and235damage pictures through four different distortion processing, in this paper, the current mainstream image criterion all have been investigated by the MOS image database of235images damage pictures, and give MOS of the the subjective evaluation.2. The main image quality evaluation method is briefly described, the mathematical model and structure diagram are presented, and the evaluation criterion of image quality evaluation of these mainstream method based on MOS image database is investigated, received a large number of experimental results, and the experimental results were analyzed and compared.3. This paper explains briefly the principle of the measurement method to measure the image quality evaluation method based on MOS image database, and gives the structure diagram. normalized parameter method to evaluate image quality has inspected based on MOS image database, so as to measure the evaluation methods,calculated the correlation coefficient of the normalized parameters and MOS in the least squares method. The inspection contains the correlation coefficient of distortiona of single image single class, distortion of single image multi class, distortion of multi image single class, distortion of multi image multi class.4. To establish the experimental platform of the correlation analysis to measure method of the image quality evaluation based on MOS image database, programming method and experimental process are described briefly, presents the experimental picture. Using the platform, the mainstream image quality evaluation method for MOS image database of different distortion types and different images are investigated, get a lot of the experimental data, and the experimental results are analyzed and compared.The method proposed in this paper to measure the image quality evaluation method based on MOS image database based on theoretical analysis and a lot of experiment. based on the experimental platform, use the main objective image quality evaluation method with all images in IVC MOS database to metric the correlation, experimental results show that the measurement method for image quality evaluation based on MOS database can measure the image quality evaluation methods quality effectively, and it can promote the improvement of the image quality evaluation methods, in favor of the discovery of the new image quality valuation method accord with human visual characteristics. |