轮廓波系数引导的数字图像修复算法研究

Research of image inpainting algorithm led by contourlet coefficients

为提高数字图像修复的精度,通过提取图像的轮廓波系数,用轮廓波系数分析图像的纹理信息,改进了基于样本的图像修复算法。引入了β因子评价图像的纹理信息,改进了优先值计算公式,优化了图像的修复顺序,使纹理复杂度较高的部分被先修复。实验结果表明,该方法有效避免了图像修复产生的图像模糊和马赛克效应,同时降低了算法的时间复杂度。

In order to improve the accuracy of image inpainting, this paper extracts the contourlet coefficients of the origi- nal image, and then uses the contourlet coefficients to analyze texture information of the image, improves the priority cal- culation formula. By introducing a factor fl to estimate texture information, it optimizes the inpainting order, repairs the complex vein parts first, and improves the sample-based image inpainting algorithm. The experimental results show that this method can avoid image blur and mosaic effect effectively, and reduce the time complexity of the algorithm.