Research on Image Fusion Method Based on Nonsubsampled Contourlet Transform

In modern military operations, as camouflage technology advances, soldiers, weaponry, and other military materials have become highly similar to their surroundings, making it difficult for traditional optical target recognition techniques to effectively identify camouflaged targets. Polarization imaging technology captures the polarization characteristics of light reflected from object surfaces, enhancing contrast between the object and the background, providing additional image information, and helping to distinguish optically similar materials or surface features, particularly excelling in complex environments. This paper proposes a multi-scale, multi-directional fusion algorithm based on NSCT, incorporating the local similarity preservation of LSP features, adaptive weight allocation, and an improved PCNN algorithm to handle details, nonlinear information, noise, and directionality issues in images. Experimental results show that this algorithm outperforms traditional fusion methods in evaluation metrics such as EN, AG, SNR, and SF, with both subjective and objective assessments demonstrating significant advantages. The proposed algorithm offers a new approach to image fusion and camouflage target recognition, demonstrating significant potential for military applications.

I. Introduction

In modern military operations, the methods for concealing soldiers, weaponry, and other military materials have become increasingly diverse and sophisticated. Camouflaged military targets have become highly similar to their surroundings in terms of color, texture, and spectral characteristics, making it difficult for conventional optical target recognition techniques to accurately identify these targets [1]. Accurately identifying enemy camouflaged targets is one of the key factors in achieving victory in military confrontations, making camouflage target recognition technology a cutting-edge research focus in recent years.