A No Reference Image Quality Assessment Metric Based On Visual Perception

With the rapid development of information technology and the widespread use of smart phones,digital images have become an important medium for people to acquire information.However,due to the transmission,environment,improper operation and differences between devices and other factors;it is easy to lead to image distortion or degradation during the image acquisition process,which gives us great difficulties to recognize the objective world and study in image.Therefore,how to accurately evaluate the quality of an image has become a hot topic in the field of image processing.Since the current no-reference image quality evaluation methods cannot accurately reflect the visual perception of human,a no-reference image quality evaluation method based on visual perception with human visual attention mechanism theory,image region division and image feature extraction is proposed in this paper,which is a general-purpose no-reference image quality evaluation method.Firstly,we deeply study the theory of the visual attention mechanism and mainly analyze the ITTI algorithm model,which successfully integrates the visual attention mechanism into the field of image processing and achieves significant results.However,there are some deviations between the extracted region of interest and human's observation.Because the texture feature and edge structure of an image are more likely to attract human's attention.Thus,this paper uses Gabor wavelet and Canny operator to extract the image texture features and edge structure features to improve the ITTI model.The experimental results show that the region of interest extracted by this method is better than the traditional ITTI and GBVS models.In addition,by comparing with the eye tracking data,the results obtained in this paper are closer to the data obtained by the eye tracker,and more in line with the results of human visual perception.Moreover,the BRISQUE method based on spatial domain is analyzed.This algorithm has the advantages of low complexity and fast execution,but it doesn't take into account the human visual characteristics.There is a certain deviation between the evaluation results of these algorithms and the subjective values.Based on this problem,this paper presents an image quality evaluation algorithm based on visual interest region.Firstly,the region of interest and the region of non-interest are divided and the statistical features of the image are extracted.Then,we merge the features and construct an objective image quality evaluation index which is consistent with the subjective feeling of the human eye.Finally,we use the evaluation index as the input of the ?-SVR,the corresponding DMOS value of image as the output,to establish the image quality evaluation model.Through testing on the LIVE2,TID2013 and CSIQ database,the experimental results show that this method has good evaluation ability in different image databases.Moreover,the performance is better than the traditional methods,which can accurately reflect the human visual perception characteristics in image quality evaluation.