An Image Quality Evaluation Based on Dense Network and Squeeze Excitation Network

In todays world., Image quality evaluation is essential in various fields such as medical imaging., security and photography since they capture high-quality visuals that are critical for accurate interpretation and decision-making. Previous researchers suggested various methods in machine learning and deep learning but still there are challenges in computational efficiency and generalizability across diverse image types and applications. This paper proposes a novel approach namely DenseNet169-SENet i.e. Squeeze Excitation Network for image quality evaluation using combination of models such as Median filter., Savitzky-Golay filter., Dense Net and SENet. Initially., the data is accumulated from a public dataset which contains X-ray images. Further., the collected data is processed through Median Filter-Savitzky Golay Filter to minimize the noise from the data and determine the size at early stage. Then., from the filtered data., optimal attributes are extracted using Dense net and., in the end., SEN et is incorporated for classification and evaluated using various metrics. The proposed DenseN et169-SEN et acquired higher results with the accuracy of DenseNet169-SENet is 97.63%., Precision is 95.77%., Recall is 98.56%., F1 score is 96.44%. when compared with existing Convolution Neural Network (CNN).