Color Image Quality Assessment Based On Quaternion Singular Value Decomposition

It is unavoidable to introduce various types and degrees noise into image when the image is processed,which has a serious impact on the acquisition of image information.Therefore,the reliable and effective objective image quality assessment technology has important research value and practical significance.The quaternion theory is applied in image processing to preserve the correlation of channels and the chroma information of images.And the base image generated by singular value decomposition(SVD)reflects the quality change of the image.This paper carries out the 2D and 3D image quality assessment algorithms based on quaternion singular value decomposition.The main contents are as follows:(1)Noise is one of the most common and varied types of distortion,but there are few studies on the noise types other than Gaussian noise.This paper proposed a non-reference color noise image quality assessment method that can evaluate five kinds of noise types without learning.The method is based on the quaternion singular value decomposition,and uses the relationship between the area enclosed by the reciprocal singular value curves of the image and the degree of the image distortion to derive a quality index.The method almost requires very little prior knowledge of any image or distortion nor any process of training.(2)Existing full-reference image quality assessment algorithms leave the contribution of image noise information to image distortion out of consideration,and are difficult to extract the noise information.Based on the conditions,this paper puts forward the image quality assessment metric based on quaternion singular value decomposition.In the proposed method,the color image and its local variance are represented by using quaternion and then performing singular value decomposition.Later,75% of singular values are taken as image noise information.We extract features from whole color images and the image noise information.Finally,these features are used as the input to a kernel extreme learning machine to predict the quality of the tested images.Since the correlation of red,green,and blue channels of color images,the characteristics of local variance,the acquisition of image noise information,and the feature combination are considered,the generated features can provide more comprehensive information from different angles.Extensive experiments demonstrate that the proposed metric has good reliability and validity and outperforms state-of-the-art image quality assessment metrics.(3)Most existing stereoscopic image quality assessment based on gray scale images lose the color information.And the 2D significant map is applied to the stereoscopic image quality assessment based on significant map,which is unreasonable and ineffective owing to ignoring the depth information and the disparity information.To solve these problems,this paper puts forward the stereo image quality assessment based on stereoscopic significant map and disparity map.In the proposed method,the 2D significance map,depth significance map and disparity map are obtained from the left view and right view of the reference stereo image pair and distorted stereo image pair,respectively.Later,the 2D significant map and the depth significant map merge into the stereoscopic significant map,which preserves the depth information and the disparity information.The left view,right view,stereoscopic significant map and disparity map are represented by using quaternion and then performing singular value decomposition to obtain the singular value and the singular vector.Then the Euclidean distances of the singular value and the SSIM scores of the singular vector are computed,respectively.Finally,the features are used as the input to a kernel extreme learning machine to predict the quality of the tested images.Experiments on LIVE 3D Phase I database and LIVE 3D Phase II database show that the proposed metric achieves high consistency with the subjective evaluations and outperforms state-of-the-art stereo image quality assessment metrics.