Infrared Image Enhancement Algorithm Based On Human Visual Model And Guided Image Filtering

Infrared image is a single-channel image that can reflect the thermal radiation information of the scene,which has unique advantages that visible images do not have,and has a wide range of applications in military and civil-related fields.However,due to the inherent defects of the imaging system itself and the influence of the complex external environment,the quality of the acquired infrared map is poor,so the infrared image needs to be enhanced.To this end,this dissertation investigates infrared image enhancement algorithms based on atmospheric scattering models,multiscale Retinex theory,and guided image filter.Firstly,an infrared image enhancement algorithm based on pseudo-dark channel prior is proposed for the problems of low contrast and blurred details in infrared images.The algorithm proposes an anti-target interference method to estimate the atmospheric illumination value,while setting a threshold lower threshold for fine transmittance to avoid image distortion,and performing feature highlighting and contrast enhancement on the reflection image of the atmospheric scattering model.Secondly,an adaptive multi-scale Retinex-based infrared image enhancement algorithm is proposed to solve the problems of detail non-accentuation and poor clarity based on the pseudo-dark channel a priori infrared image enhancement algorithm.It mainly proposes an adaptive multiscale weighting function based on local variance to linearly weight the reflection components to highlight local detail information.Further,using nonlinear correction and adaptive stretching to process the texture structure hidden in low gray areas.Finally,to address the issue of the adaptive multi-scale Retinex does not suppress the noise when enhancing the image details,an infrared image enhancement algorithm based on the detail enhancement guided image filter is proposed.This algorithm proposes a detail enhancement guided image filter by designing edge perception factor and detail regulation factor.Further,the detail regulation factor is introduced into the detail layer to enhance the suppression of flat region noise,and fuses the enhanced detail layer with the base layer.In this dissertation,we conduct simulation and comparison experiments of the above three algorithms in multiple scenarios through public datasets.The experimental results show that the algorithm has certain advantages in both subjective and objective evaluations,and lay the theoretical and technical foundation for the realization of a high-performance infrared imaging system.