Super-Resolution Reconstruction Of CCD(CMOS) Image Based On Laster Active Long-range Imaging

High-resolution imaging for long-range active laser detection have been widely used in many applications,including military defense field,socioeconomic field and other fields.However,the detection limit of traditional photo detectors and the extreme attenuation of mixed returned echo signal has become the bottleneck of the development of the long-range active laser detection.It is difficult to continue to improve the detection space resolution from the hardware,which brings great challenges to long-range active imaging.With the optical system and photoreactor performance basically unchanged,the super-resolution reconstruction technology can carry out high-resolution image reconstruction without optimizing the hardware system,which has a significant advantage,which has significant advantages.Therefore,the use of multi-frame sequence low-resolution image information for super-resolution reconstruction to improve image has attracted widespread attention from academia.This paper is an attempt to perform super-resolution reconstruction based on the optimal estimation model of genetic algorithm.A new technology for achieving better reconstruction results.This algorithm is relatively simple in the calculation model.It is easy to introduce prior information,which has better robustness.The considerable research efforts have been devoted to super-resolution reconstruction on the known high-resolution images after degradation,indicating the role for genetic algorithm.However,Super-resolution reconstruction technology for different kinds of image had been a largely under explored domain,especially for laser long-range active detection imaging.The research of superresolution reconstruction technology based on the genetic algorithm,at present,is mainly to simulate the reconstruction of known high-resolution images after degradation,to verify the effectiveness of the genetic algorithm in the application of super-resolution technology.The author's pioneer work has contributed to super-resolution reconstruction technical application.The method is found by applying the genetic algorithm to the target image resolution improvement for the practical application of actual laser long-range detection,which increases the spatial sampling rate by time-domain sampling rate.Super-resolution on CCD(CMOS)image of laser long-range detection has briefly introduced in this paper,which has started from the genetic algorithm(GA)of the optimum estimate computational models in the solution space.The theoretical basis of superresolution imaging and existing technologies are theoretically studied has been given in the first section of this paper,including image super-resolution reconstruction rationale,image super-resolution reconstruction method and evaluation criteria.Further discussion of the characteristics of atmospheric transmission has been given in the second section of this paper,including atmospheric scattering,atmospheric absorption,atmospheric radiation,atmospheric turbulence and other effects on the laser long-range active detection.A study deals with the sub-pixel scanning information brought by atmospheric disturbance and selfscanning in laser long-range imaging,which sub-pixel motion estimation and alignment by block matching method.This high-resolution method which is based on the genetic algorithm that mimics the biological group's natural adaptation to the evolutionary process and the genetic diffraction mechanism model.The search process of guided solution space according to probability gradually converges in the direction of optimal solution to complete the optimization estimate,so as to realize the super-resolution reconstruction of laser longrange detection image.An improved strategy to solve this the problems of “precocious” and “ringing” in the process of image reconstruction problem is presented,by using the gradient information of the image itself.The new method deals with the fitness function on prior information such as the continuity and variability of pixel values,which evaluates the similarity of image edge features and the gray variance of image self-similar regions as constraints.In addition,an approach to accelerate the convergence speed of global optimal estimation of genetic algorithm is presented,which introduces adaptive linear search process to improve the local search capability.The experimental results of super-resolution imaging with a detection distance of 5km indicate that the super-resolution reconstructed image based on the improved genetic algorithm for long-distance detection of laser has clearer edge details.However,the evaluation indicators such as standard peak signal-to-noise ratio and structural similarity are greatly improved,which has a certain suppression effect on speckle noise generated by laser remote detection.It has more obvious advantages in image reconstruction speed and quality.