High Dynamic Range Infrared Image Detail Enhancement Based On Visual Characteristics And FPGA Implementation Technology

By passively receiving infrared radiation to obtain scene information,infrared imaging system has advantages of good concealment and working day and night without being affected by time and light.It is widely used in different fields such as military and civil fields.But at the same time,infrared images usually have characteristics of high dynamic range,low contrast and low signal-to-noise ratio,which seriously reduces image quality and visual effects.In addition,due to high dynamic range,infrared images cannot be directly displayed on traditional display devices.For better observing and monitoring,it is necessary to compress the dynamic range of infrared images and enhance the details.Since the human visual system has the most advanced scene perception and information processing capabilities,image enhancement technologies based on human visual characteristics have attracted more and more attention in recent years.Based on the visual characteristics of the human,a visual attention distribution map will be obtained in this thesis through the visual saliency model to guide the adaptive detail enhancement of infrared images.In this thesis,the perceptual characteristics of human visual system are analyzed firstly.This thesis focuses on several visual saliency algorithms,and then a multi-scale local contrast saliency algorithm based on the mean of neighborhood is studied to meet the need for better guidance on infrared image detail enhancement.And the simulation results have verified the effectiveness of the algorithm.Secondly,in view of the problems of gradient inversion and halo in the present image enhancement algorithms based on unsharp mask,several typical algorithms are analyzed in this thesis,such as guided filtering,weighted guided filtering and side window guided filtering algorithms.On this basis,a side window weighted guided filtering algorithm based on local differences is proposed.The experimental results have demonstrated that the proposed enhancement method can effectively smooth the details while improving the edge retention characteristics of guided filter,avoiding over-enhancement of strong edges.Aiming at the problems of blurred target details and over enhanced background in traditional infrared image enhancement algorithms,an infrared image enhancement algorithm based on visual characteristics and improved guided filtering is proposed in this thesis.In this algorithm,firstly,an infrared image is divided into a base layer and a detail layer by the side window weighted guided filtering algorithm based on the local difference,then the visual saliency map is used to guide the adaptive enhancement of the detail layer,while the Sigmoid curve based on the effective extreme values is used in the dynamic compression of the base layer.Finally,the processed detail layer and base layer are fused together to obtain the final enhanced infrared image with low dynamic range.The experimental results have shown that the proposed algorithm has better regional selectivity,which can enhance target details effectively while suppressing noises.The enhanced infrared image can better meet the characteristics of visual perception.For real-time implementation of the infrared image enhancement algorithm,a hardware system based on FPGA is designed in this thesis.And the principle of the implementation is introduced in detail.Finally,the performance analysis and resource evaluation results of the system have verified the superiority of the system.In additon,the system can meet the expected goals of real-time realization,and has better portability and larger upgradeable space.