Image Noise And Blocking Artifact Assessment

Noise is introduced into image in multiple parts such as acquisition,storage,transmission.Image noise has a great influence on the acquisition of image information.In addition,the prior experience of noise distribution in many image processing algorithms is also important.If the algorithm parameters can be set according to the prior experience,it will be very helpful to achieve the desired algorithm results.Blocking artifact of image mainly refers to the discontinuity between restored image blocks caused by block discrete cosine transform in the process of image compression and coding.The existence of image blocking artifact also affects the acquisition of image information.If the blocking artifact of the image can be accurately evaluated,the compression ratio can be as high as possible to ensure that the image quality can be accepted by the end user,so that the transmission speed of the image data can be improved and the storage cost can be reduced.Therefore,accurate assessment of blocking artifact and noise in images is of great significance to image processing algorithms.Image noise assessment is also called image noise level estimation.The traditional image noise assessment algorithms greatly reduce the accuracy of the evaluation results when the texture information of the image is rich or the content changes greatly.To solve this problem,a piecewise noise level estimation method based on BM3 D is proposed.Firstly,a simple and fast noise estimation algorithm is used to estimate the noise level of image roughly,and the preliminary estimation value is used as the guidance to generate multiple candidate noise level values.Then,the candidate noise level values are used as the filtering parameters to perform multiple BM3 D filtering of noise image.It is proved that the difference of NSS features before and after filtering can reflect the efficiency of image denoising.On the basis of the results,the distance between the NSS feature vectors of the filtered image and the noise image is taken as the weight of the candidate value of the noise level corresponding to each filtered image,and the final accurate noise level is calculated by using the local mean method.The experimental results show that the accuracy and reliability of the proposed method are greatly improved compared with the existing noise level estimation algorithms.In view of the shortcomings of traditional blocking artifact assessment in combining the relevant characteristics of HVS model,a blocking artifact assessment method which combines image gradient and human eye masking effect is proposed.Firstly,based on the gradient information of the pixels at the edge of the image block,the image blocking artifact map is obtained,which mainly including the location and intensity information of the image blocking artifact.Then,the luminance and texture masking effect of the human eye on the image are calculated and combined into the blocking artifact map to obtain the noticeable blocking artifact map.Finally,the statistical method is used to calculate the noticeable blocking artifact map,and the evaluation index of image blocking artifact is obtained.The experimental results show that compared with the existing blocking artifact assessment algorithms,the proposed algorithm has the highest monotonic consistency between the evaluation index of the test image and the human subjective evaluation index.