Research on Image Fusion Technology Based on Compressed Sensing

In the field of image processing, the development of compressed sensing technology has attracted widespread attention. In this context, research on compressed sensing-based image fusion techniques has also gained increasing importance. To better process image information, a novel algorithm—compressed sensing domain image fusion based on high-frequency and low-frequency importance metrics—is proposed. This algorithm initially obtains measurements using a dual-star sampling mode (implemented via structured sampling matrices), then computes importance metrics for high- and low-frequency regions as fusion operators (using energy-based weighting functions), and performs weighted fusion of the measurements. Finally, the fused image is reconstructed by solving a minimum total variation optimization problem (e.g., using gradient descent or ADMM algorithms).

To evaluate the algorithm’s performance, a series of subjective and objective experiments were conducted. The results indicate that the algorithm excels in image fusion, surpassing other Fourier-based schemes. Through this new compressed sensing domain image fusion approach, image information can be processed more effectively, enabling better applications in image processing and computer vision domains.