Year-end Sale % OFF 
Abstract
In order to overcome the problems such as poor global search ability, slow convergence rate, and easy to fall into local minimum values in the image denoising process of traditional BP neural networks, the HS-LMBP hybrid neural network image denoising algorithm is proposed which combines the harmony search algorithm and the LMBP algorithm. The HS-LMBP hybrid neural network algorithm combines the high speed of the LMBP algorithm and the global nature of the HS algorithm, which can be a good improvement to the existing problems of the BP algorithm model. Compared with the Wiener filtering, BP, LMBP and PSO-LMBP model image denoising effects, the denoising model using HS-LMBP neural network algorithm has a better denoising effect.
Highlights
With the rapid development of modern information technology, images, as an important information carrier for people to understand the objective world, are exerting more and more important influence on people's lives.Image denoising is one of the classical problems in image processing research
Step6.Optimal harmony decoding----The Back Propagation (BP) neural network initial weight threshold w, v, r are obtained by decoding the Harmony search (HS)-optimized solution values corresponding to the optimal harmony, and theyare given to the neural network image compression model based on the LMBP algorithm for secondary training, perform secondary training until the error function E(x(k) ) goal calculated according to (1) or the maximum number of operations epochs
Using Gaussian Noise Image with 150×150 Brain and Cameraman as Test Samples,Wiener filtering, BP, LMBP, PSO-LMBP, and HS-LMBP algorithms are used to model denoising experiments respectively.Figure 6-12 shows the results of the relevant denoising experiments on the Brain chart. It can be seen from the figure that the Winner filtering does not remove the noise well and the image quality is not significantly improved compared to the noise-containing graph.The BP algorithm model is unable to eliminate the noise and makes the image details more blurred due to the limited number of operations into the local optimum, while the LMBP, PSO-LMBP and HS-LMBP have achieved a good denoising effect.Similar results were obtained from the related experiments of Cameraman diagrams in Figure 13-19.In order to further quantitatively evaluate several algorithms, structural similarity (SSIM) functions and peak signal-to-noise ratio (PSNR) functions are used as evaluation criteria
Summary
With the rapid development of modern information technology, images, as an important information carrier for people to understand the objective world, are exerting more and more important influence on people's lives.Image denoising is one of the classical problems in image processing research. The denoising effect of these methods is good, but the point spread function is difficult to solve, so this type of method The practical application range is limited.The opposite is the image blind denoising method, which can estimate the effective point spread function without any prior information and solve the problem. This kind of blind denoising method is widely used at present, but there is a large amount of calculation, local convergence, and the solution is not the only problem[1]. Because the BP algorithm is based on gradient descent principle, there is a weak global search capability and slow convergence speed. , easy to fall into local minimums and other issues.In order to overcome these problems, the improved BP algorithm (Levenberg Marquardt, LM) algorithm is often used to build model ,This algorithm combines the local convergence of the Gauss-Newton method and the global convergence of the gradient descent method,the efficiency of the LMBP model for image denoising is significantly better than the gradient descent BP algorithm.because it can't fundamentally solve the problem of local minima caused by the randomness of the initial weight and threshold of the BP algorithm, the model optimization effect is limited.Harmony search (HS) algorithm is a novel intelligent optimization algorithm proposed by Korean scholar Geem Z W et al in 2001 [2],the existing research results show that this algorithm can well solve the problem that the BP algorithm is easy to fall into the local minimum value caused by improper choice of initial and weight threshold [3].a HS-LMBP mixed-image denoising algorithm based on the harmony search algorithm and LMBP algorithm is proposed,experimental results show that compared with Wiener filtering, BP, LMBP, PSO-LMBP algorithm model image denoising effect, using HS-LMBP algorithm modeling image denoising can achieve better results
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
