Wavelet Inpainting Model Driven by Curvature ★

In the total variation wavelet inpainting model, when the low-frequency wavelet coefficients and almost all high-frequency wavelet coefficients, in some areas, are all lost, the wavelet coefficients always can not be filled properly. Based on the defect, the curvature term is introduced into the model to guide the filling process, the Euler-Lagrange equation of the new model is derived through variational method. In the image domain, the same gray level line can be jointed under the guide of its curvature, then the wavelet coefficients can be filled as more as possible, the repaired image will be more faithful to the initial one. Experiment results show that the lost wavelet coefficients can be filled more efficiently through our new method.