Efficient 2D and 3D Image Classification and Compression Using Dual-Hahn Moments

This paper focuses on improving the Dual-Hahn moments to boost the efficiency and accuracy in tasks such as compressing and classifying high-quality 3D and 2D images. The proposed method, which uses the Deep Neural Network algorithm to classify images based on Dual-Hahn moments as features in first layer, seeks to precisely characterize the image by representing it as a vector called the descriptor vector of moments. This vector is subsequently utilized to reconstruct the image for each order. The simulation and analysis of the results illustrate the superior quality and efficiency of the proposed DHM-DNN method for image compression, and classification, as well as its exceptional accuracy in both noisy and noise-free images.