Research on Dynamic Weight Feature Extraction Based on Gray-Level Distribution

Currently, the majority of published methods for dynamic image feature recognition primarily rely on rough extraction of color co-occurrence matrices. Consequently, these methods are unable to effectively detect features in images with substantial dynamic range variations and extreme interference. To address this issue, this paper proposes a dynamic weight-based feature enhancement extraction method for color images, grounded in grayscale distribution. The technique utilizes a spatial grayscale distribution as the core module, it introduces dynamic weight allocation for depth variation, along with finite element optimal solution algorithms to enhance multi-level feature weight integration. By comparing the extracted features with the original grayscale distribution, the algorithm demonstrates robustness and precision. Extensive comparative experiments have been conducted, and the results indicate that the proposed method significantly improves the performance of image feature extraction, achieving the best results to date, with a 10% increase in accuracy compared to traditional feature extraction methods. Additionally, ablation experiments have been performed to validate the rationality of the proposed model's structure.