Underwater Image Restoration Using No Template Model Tracking Method Based On Improved Bat Algorithm

Digital image restoration technology has achieved rapid development in recent years,but image restoration for underwater scenes has been relatively rare.This is due to the complicated and varied underwater optical environment and disturbances in the water surface.Distortion of underwater images caused by different reasons requires different methods to recover.This paper studies the restoration of underwater distorted images from two perspectives: water degradation images caused by light attenuation and underwater distortion images caused by water surface distortion.In this paper,we found that the traditional non-physical model restoration method does not work for the water degradation image caused by light attenuation.Therefore,a water degradation image restoration method based on image wavelet transform and homomorphic filter fusion is proposed,and the water decline image is passed through the wavelet.The low-frequency image is decomposed and preserved,and the low-frequency image after homomorphic filtering and the three high-frequency images of the original image horizontal detail image,vertical detail image,and diagonal detail image are reconstructed by wavelet to obtain a restored image.Experimental results show that the method greatly reduces water decrease the average gray value of the quality image,and less image distortion.For underwater distorted images caused by water distortion,the topological changes in water surface fluctuations are characterized by high-speed motion,randomness,and extremely high complexity.The accurate water surface shape is obtained,and the real scene that recovers the bottom of the water through distorted images is The key to achieving underwater distortion image recovery.This paper introduces a method for underwater distortion image restoration: no template model tracking method.Since the objective function of non-template model tracking method is non-convex function,the traditional optimization algorithm may fall into a local optimum when solving extreme values.In order to get a clearer restoration image,this paper proposes an adaptive differential evolution bat algorithm(ADEBA),which combines the mutation operation,crossover operation and selection operation in adaptive differential evolution with the basic bat algorithm to improve the basic bat algorithm.It is trapped in the problem of local extremum,and due to the adaptability,the convergence accuracy and convergence speed of the algorithm are balanced.The result of the test function shows that the optimization ability of the algorithm proposed in this paper is greatly improved.Finally,the excellent function of ADEBA is used to optimize the objective function of the no template model tracking algorithm,and then a more accurate distortion function is obtained,and finally a clearer result is obtained.Compared with other classic bionic optimization algorithms,the underwater image restoration is better based on the ADEBA no template model tracking method,and the image quality is higher.