Mazin M. Image compression methods in Internet of Things systems on microcontrollers

The dissertation is devoted to solving a topical problem - studying the effectiveness of using wavelet transforms for image compression in embedded systems with limited resources and determining the optimal compression parameters. For this purpose, existing compression methods were analysed, their comparison was carried out, the main shortcomings were identified and ways to eliminate them were proposed. The work is the first to set and solve the problem of finding the influence of the decomposition level of the wavelet transform on the quality of the restored image in microcontroller systems, and to study the features of the operation of different types of wavelets (Haar, Daubéchy, Coiflet) in combination with JPEG compression. For the first time, a comparative analysis of the use of entropy compression algorithms, in particular the Huffman and RLE methods, in the context of their application for limited computing resources was carried out. A new method of adaptive correction of artifacts that arise during repeated use of wavelet transforms in combination with JPEG compression was proposed and implemented, which allows improving the quality of the restored image without a significant increase in computational costs.