The Research Of Digital Image Denosing Model

This paper mainly researched the denoising problems in the field of imageprocessing. By using the variational method, PDE base image denoising model. Thebasic idea of the variational method: by using the variational method, the denoisingproblem is converted to energy functional minimization problem, derived theEuler-Lagrange equation, using the negative gradient flow method to obtain theevolution equation, the numerical experiments give the results of the model.The mian research results are as follows:1. An image denoising model based on feature detection functionwe made a detail analysis to image denoising model-total variation model (ROFmodel), based on it we propose a feature detect function denoising model to defend theblock effect in the ROF model, using feature detecting function to detect the imagetexture information to balance the regularization term and the fidelity term, whichmakes the model can both denoising and maintain the texture information of the picture.The new model can reduce the block effect, the restored image is closer to the originalimage.2. A local variance adaptive denoising modelStudying the multiplicative noise model-RLO model, we propose a local varianceand self adjusting model based on RLO model, it can protect detail information fromdenosing. We made numerical experiments on both the two models, the experimentsresults show that our new models do well in reducing the blocking effect and preservingimage details.3. A new iterative denoising algorithmBy studying the additive noise denoising model and the multiplicative imagedenoising model, we propose a new alternating iterative denoising (AID) algorithmcombining the TV-L1model to remove additive noise with the AA model to wipe offmultiplicative noise. We make experiments on additive noise color image andmultiplicative color image respectively. The simulation results show that the AIDalgorithm is effective, stable and rapid.