Adaptive vascular enhancement of flap images in the second near-infrared window based on multiscale fusion and local visual saliency

Near-infrared fluorescence imaging in the second window has emerged as a valuable tool for the non-invasive and real-time assessment of vascular information in skin flaps. Enhancing flap images to provide more accurate flap vascularization information is critical for predicting flap viability. To address the limitations of existing methods in enhancing vessel images, we propose a novel and adaptive technique for enhancing flap microvessel images. Multiple strategies can be employed to effectively enhance the visualization of small-scale vessels. Firstly, the proposed method leverages the multiscale rolling guided filter to acquire the base layer and detail layers at different scales. Furthermore, correlation coefficients are utilized to weigh and fuse the detail layers effectively. To suppress noise amplification while enhancing vascular structures, an improved adaptive gamma correction method based on local visual saliency is introduced. Meanwhile, the bilateral gamma correction is used to enhance the base layer. Finally, the enhanced base layer and detail layer are fused using the weighted fusion strategy. We conducted experiments on skin flap vessel images, retinal fundus images, finger vein images, and low-light images. Our method achieved excellent results in metrics such as NIQE, AMBE, and WPSNR, demonstrating significant advantages in preserving the structural integrity and brightness consistency of the images. The obtained results validate the potential of this method in enhancing vascular images, indicating promising prospects in the field of medicine.