基于DCT同态滤波的纸病图像去噪算法研究

Paper Defects Image De-noising Algorithm Based on Holomorphic Filtering and Discrete Cosine Transform

关键词:  DCT变换  同态滤波  高斯滤波  频率域

Key Words:discrete cosine transform （DCT）  holomorphic filtering  gaussian filtering  frequency domain

摘要:针对纸病图像采集过程中容易受到光源噪声干扰的问题，在分析了光源噪声特性的基础上，提出基于DCT同态滤波的纸病去噪算法。该算法首先对采集到的纸病图像进行分块，然后对每一图像块取对数，DCT变换，将其变换到频率域，使用高斯同态滤波滤除噪声干扰，最后通过IDCT变换以及取指数操作将其恢复到空间域。实验结果表明，该算法可以有效地去除纸病图像中的各类光源噪声干扰，增强纸病图像区域。

Abstract:Aiming at the problems of the light-source noise in the process of paper defeats collection，the paper analyzed the characteristics of the light source noise and proposes a de-noising Algorithm based on holomorphic filtering and DCT transform. Firstly， blocking the image; then， taking logarithm operation and DCT transform of each image block and transforming them to frequency domain. Next， Gaussian holomorphic filteringused to remove noise interference. Finally， IDCT transform and exponential operation are used to restore them to the spatial domain. The experimental results showed that the algorithm could effectively remove various kinds of light-source noise and enhance the paper defect area．