Image Quality Assessment Algorithm Research

Image quality assessment is an index which measures image quality is good or bad. Image quality assessment plays a key role of accessing the information of the image fully and accurately. Studies have shown that the edge structure information of image and the low-level features of image are important for image quality assessment. In the field of image processing, the main information of some images whose information is only concentrated on a few key areas, and another which information is concentrated on the whole image and the requirements of whether conform with the human visual system is very high. So the article will focus on these two types of images and combine the edge structure information and the low-level features of image in two aspects:on the one hand, combined with the interested region as well as the image edge information proposed the edge structure similarity quality algorithm which is based on the interesting region, on the other hand, combined with an image quality evaluation of whole image, the image edge information, and the bottom of the image feature similarities proposed image quality assessment of feature similarity combined with gradient information.Full reference image quality evaluation became the most commonly used image quality assessment algorithm because of its simplicity and accuracy in this time. The objective image quality assessment algorithm--SSIM algorithm with simple and efficient characteristics can reflect the human visual system in a good way, so it has been used more widely, but the algorithm didn’t bring out the edge feature which was taking along important information in the image, and it has a certain error when it evaluates the quality of image. Though ROI and dual-scale edge structure similarity algorithm has taken the importance of edge information into consideration, this algorithm is not an ideal one in the identification of edge information. Based on the above information, this article proposed the edge structure similarity quality algorithm which is based on the interesting region. This algorithm departs the image into interesting and non-interesting region, and evaluates the interesting region with the edge structure similarity quality algorithm. The experimental results showed that this algorithm has a stronger ability to identify the edge Information and is more sensitive to the variations of the image quality.Image quality evaluation method of FSIM based on the low-level features has taken the importance of the low-level features into consideration, but this algorithm is not an ideal one in the identification of edge information. Based on the above information, this article combines FSIM algorithm with GSSIM algorithm which is more sensitive to the edge information so as to get a new image quality assessment FGSIM which is not only consistent with the human visual system characteristics, but also can effectively identify the image edge. The algorithm combines the part of FSIM algorithm which represents phase consistency with the part of GSSIM algorithm which can extract image information so as to get a new image quality assessment FGSIM. Among them, Using phase consistency represents image features, the part of phase consistency can be used to keep the algorithm close to the human visual system, and the part of GSSIM algorithm which can extract image information realized by Gradient, the part can be used to identify the image edge. The experimental results show that the algorithm is not only has a stronger ability to identify the edge Information, but also is more sensitive to the variations of the image quality.