Integrating CNN and Random Forest Algorithm for Multi Satellite Image Compression in Data Mining

At the moment, scholarly attention is being drawn to a significant issue called multispectral image compression (MSIC). Radar systems, satellite communications, and sensing technologies are always keeping an eye on the environment, space, and Earth. Resources like power consumption and storage are becoming more and more affordable in the competitive marketplace. Moreover, performance capability is still limited. Multispectral image processing is required during this process. The methods and the conditions for using them must be followed. Satellite cameras keep an eye on related optical data, water wells, disaster surveillance, and geographic information. Machine learning statistical parameters, attacks, histogram equalisation, and compression of image quality all require improvements. Existing approaches mostly use low-rank tensor decomposition, DWT segmentation methods, and matrix-based modeling; nevertheless, they fail to distinguish between different strip components and have several drawbacks. Moreover, problems with spectral redundancy and the removal of sub-band models can be resolved using machine learning. The Natural Random Forest Machine Learning Model (NRFML) is utilized in this study. This model trains and compresses multispectral images for a variety of uses. In the end, it contrasts MSE, PSNR, NCC, and SSIM metrics with current techniques and finds that the proposed NRFML is competitive with past methods.