Abstract
An image quality assessment (IQA) aims for predicting the quality of the images with or without any previous information of reference image. Assessing image quality is significant in image communication and processing. There are several techniques been proposed for quality assessment, but the accurate feature learning and complexity analysis are still challenging issues in various image processing applications. To improve the quality assessment with minimum complexity, an adaptive frost filtered quantile regression-based artificial deep structure learning (AFFQR-ADSL) framework is introduced. The AFFQR-ADSL framework performs the image quality assessment based on two methods, namely full reference and no reference. The AFFQR-ADSL framework with full reference (AFFQR-ADSL-FR) is carried out based on the test image with the reference image information. Initially, the numbers of input images and reference images are collected from the image database. The proposed AFFQR-ADSL-FR framework trained the input images and reference images with different layers to progressively learn the higher-level features from the raw input. The numbers of images are given to the input layer of the artificial feed-forward deep structure learning network. Then, the inputs are transferred into the hidden layers to repeatedly learn the features. In the first hidden layer, the input images are preprocessed to filter the noise present in the input images using an adaptive mean frost filtering technique. Followed by, the input images are divided into the number of patches and feature extraction is carried out in the next successive hidden layer. Then, the learned features are combined and fed into the output layer. Finally, the proposed technique uses linear quantile regression at the output layer for analyzing the extracted feature vectors and obtains the quality assessment results. Then, the AFFQR-ADSL framework with no reference (AFFQR-ADSL-NR) is carried out by using the test image without using the reference image information, extracts the feature vectors, and provides the assessment results at the output layer. Result evaluation is carried out with metrics such as mean square error, peak signal-to-noise ratio (PSNR), accuracy, and computational complexity (CC). The qualitative and quantitative results show that the proposed AFFQR-ADSL-FR and no reference achieve better results in terms of PSNR, error, and CC.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.