RECURSIVE WAVELET TRANSFORM ALGORITHM FOR IMAGE PROCESSING

Abstract

This study presents the design and real-world implementation of advanced recursive image compression algorithms based on wavelet transform techniques. The proposed method aims to improve overall compression efficiency while preserving high image quality and optimizing computational performance. Тhe visual quality of reconstructed images while simultaneously reducing computational demands to accelerate processing speed. A multilevel recursive wavelet decomposition scheme is formulated to provide adaptive, scale-sensitive analysis of image features. To improve the segmentation stage, a unified quadtree-based partitioning strategy is proposed, enabling flexible subdivision of image areas according to local texture characteristics and intensity distribution. Furthermore, the study presents a parametric segmentation method that relies on ranking block intersections. Candidate domain blocks are assessed using similarity measures and ordered through a prioritization mechanism, which significantly limits unnecessary comparisons and reduces the effective search space. This optimization decreases the total number of domain blocks involved in encoding, thereby improving computational efficiency without negatively affecting compression effectiveness.The study was conducted using a dataset comprising 102 flower categories, each containing at least 40 images. The photographs were captured from multiple perspectives and under varying conditions to ensure diversity and robustness of the data. lighting conditions to ensure variability. Performance evaluation considered objective similarity indicators, encoding time, and overall processing performance. Although the compression ratios achieved by all methods were comparable, examined models were nearly equivalent, noticeable differences were observed in processing speed. Among the deep learning architectures incorporated into the preprocessing stage, MobileNet V2 demonstrated the highest computational efficiency while maintaining strong reconstruction accuracy. As an outcome of this research, a dedicated software application was created to realize the recursive compression framework built upon wavelet transform techniques. The application provides fast and reliable image data compression by integrating content-independent compression techniques with adaptive segmentation procedures, making it well suited for large-scale image processing applications.