Measuring Image Quality: Crucial Indicators for Evaluating Camera-Captured Document Contents

In this research, we assess the effectiveness of six non-reference quality measures for image quality evaluation in the context of distributing multimedia content and medical images.: BRISQUE, NIQE, PIQE, MLV, DCF, and a hybrid of BRISQUE and MLV. These measurements have been put into practice using Python technology and evaluated on a dataset that is openly accessible. Additionally, we analyse the effectiveness of the major classification algorithms—KNN, SVM, Decision Tree, and Naive Bayes—as well as the Random Forest Ensemble Method for analysing measurements and forecasting the image quality for predicting picture quality and doing measurement analysis. We evaluate each classifier's performance for measurement analysis and picture quality prediction by comparing its accuracy, precision, recall, and F1 score. Our study's findings indicate that the best connection with subjective quality scores for both photos is obtained when BRISQUE and MLV are combined. The Random Forest method performs the best, ranking second in terms of classification accuracy and F1-score, after SVM and Decision Tree. Our findings show that these quality measurements and classification models have a lot of promise for application in real-world picture quality evaluation jobs.