论文部分内容阅读
提出一种基于各向异性扩散偏微分方程的红外图像噪声抑制算法。通过将形态学处理和红外图像局部特征相结合,建立了一种新的扩散系数。该系数利用形态学膨胀腐蚀操作获取梯度算子,改善了Perona-Malik(P-M)梯度算子对噪声的敏感性,实现了均匀区域扩散增强且边缘细节区域扩散减弱的目的。算法已在EVM-DM642硬件平台上实时运行,实验表明:它在有效平滑噪声的同时较好的保持了图像边缘细节信息。
An infrared image noise suppression algorithm based on anisotropic diffusion partial differential equation is proposed. By combining the morphological processing with the local features of infrared images, a new diffusion coefficient was established. This coefficient uses the morphological swelling operation to obtain the gradient operator, improves the sensitivity of the Perona-Malik (P-M) gradient operator to noise, and achieves the purpose of enhancing uniform diffusion and weakening of the edge detail region. The algorithm has been run in real-time on the EVM-DM642 hardware platform. Experiments show that the algorithm preserves the image edge details while effectively smoothening the noise.