Positive Assessment Of Image Quality Based On Artificial Intelligence

As two-dimensional signal,image can intuitively transmit information to human brain.However,there are many disturbances in the process of image formation and storage,which is easily resulting in the loss of information in the image and the decline of visual quality.Therefore,it is necessary to establish a corresponding evaluation mechanism to evaluate the quality of the image.And the advent of the era of intelligence makes it possible to replace the human brain with machine to evaluate the image quality.This thesis mainly evaluates the image quality from the following two aspects.Aiming at the problem that the current no-reference image quality assessment algorithm lacks of enough accuracy to predict image quality,a no-reference image quality assessment method based on spatial domain coding is proposed.This method firstly extracts the spatial structure features of the image on different bit planes.These features can quantify the structure information between the pixels to reflect the distortion degree of the image more accurately.Then,an image quality assessment model is established by neural network.The experimental results in LIVE,CSIQ and TID2013 databases show that the proposed algorithm is more accurate than the existing mainstream image quality assessment algorithms,and has high consistentence with the subjective perception of human eyes.Finally,an automatic focusing method based on the quality assessment algorithm is designed and applied to the automatic focusing of camera,which verifies the validity and practicability of the algorithm.And to deal the problem that it is difficult to accurately and effectively extract the quality features of mixed distortion image,an image quality assessment method based on spatial distribution analysis is extracted.In this method,the brightness coefficients of the image are normalized firstly,and the image is divided into blocks.While the CNN is used for end-to-end depth learning,the multi-level stacking of convolution cores is applied to acquire image quality perception features.The feature is mapped to the mass fraction of the image block through the full connection layer,then the quality pool is obtained by aggregating the quality of the block.Through the analysis of the spatial distribution of local quality in the quality pool,the features that can represent its spatial distribution are extracted,and then the mapping model from local quality to overall quality is established by the neural network to aggregate the local quality of the image.Finally,the effectiveness of the algorithm is verified by the performance tests in MLIVE,MDID2013 and MDID2016 mixed distortion image databases.