光学图像的自相似分块增强研究

Optical image enhancement has a significant research value in optical information processing and measurement. In order to adapt to the complex structure of optical images, based on the ideas of image self-similarity and block processing, using the technologies of the self-similarity processing, the shock filter with anisotropic diffusion, the sparse representation, the flux limit technique and adaptive processing to image features, we propose two methods of optical image enhancement using image self-similarity and block processing. (1) For general optical images with additive noise, a feature adaptive denoising and sharpening with a balance between the smoothing and the sharpening operators; (2) for ESPI images in optical measurement with both additive and multiplicative noise, a joint sparse denoising method after a grouping of blocks with self-similarity in logarithmic domain. At the same time, we study their modeling, model analysis and fast efficient numerical methods. As a cutting-edge cross-over study involving information and mathematics, this project has important theoretical and practical value, which will further deepen the applications of variational methods and partial differential equations, the self-similarity processing and the sparse repre- sentation theory in image enhancement.

光学图像增强在光学信息处理和测量中具有十分重要的研究价值. 为了适应光学图像复杂的结构, 基于图像的自相似性和分块处理思想, 利用自相似性处理, 冲击扩散方程, 稀疏表示, 通量限制技术和特征自适应处理等技术, 提出两个光学图像的自相似分块增强处理方法. (1) 对于带有加性噪声的一般光学图像, 特征自适应的去噪和锐化方法; (2) 对于带有加性和乘性噪声的电子散斑干涉条纹图像, 对数域中自相似块分组的联合稀疏去噪方法．同时, 对于它们的建模、模型分析和快速高效的数值实现方法进行研究．作为涉及信息与数学的前沿性的交叉研究, 本课题将进一步深化变分偏微分方程方法、自相似性处理技术和图像稀疏表示理论在图像增强中的应用, 具有重要的理论和应用价值．

光学图像增强在光学信息处理和测量中具有十分重要的研究价值. 为了适应光学图像复杂的结构, 基于图像的自相似性和分块处理思想, 利用自相似性处理, 冲击扩散方程, 稀疏表示, 通量限制技术和特征自适应处理等技术, 重点研究了光学条纹模式图像的自相似分块增强处理方法. 在掌纹识别、视网膜血管增强、低剂量CT成像、干涉条纹滤波和光学图像增强等方面提出了若干国际先进的光学图像增强算法, 得到了较好的图像处理结果. 作为涉及信息与数学的前沿性的交叉研究, 本课题进一步深化了变分偏微分方程方法、自相似性处理技术和图像稀疏表示理论在光学图像增强中的应用, 在医学影像处理、光学测量和智能视觉计算等领域具有重要的理论和应用价值．