基于小波变换和高斯差分冷冻电镜生物大分子图像的自动分割

冷冻电镜生物大分子图像分割是进行冷冻电镜生物大分子颗粒识别的基础.文章分析了冷冻电镜生物大分子图像的主要特点,提出了基于小波变换和高斯差分的冷冻电镜生物大分子图像自动分割方法.该方法利用小波变换得到原图像的低分辨率图像,抑制了噪声,提高了图像的对比度;同时采用高斯差分算子解决了图像亮度不均匀的问题,并对高斯差分图像采用基于灰度梯度信息融合的分割方法.实验结果表明,该算法能有效的减少噪声对边缘提取的影响,分割效果良好,是一种全新的冷冻电镜生物大分子图像自动分割算法.

Automatic segmentation for cryo-electron micrograph image based on wavelet transformation and difference of Gaussian

Segmentation of cryo-electron micrograph (Cryo-EM) image is crucial to identification of macromolecular particles. In this paper, the features of Cryo-EM image are discussed and a new automatic segmentation algorithm for Cryo-EM image based on wavelet transformation and difference of Gaussian (DoG) is also provided. This method uses wavelet decomposition to do the multi-resolution analysis, which can depress noise and enhance contrast. After DoG image is segmented by using OTSU method, main edges are selected by the combination of gray and grads information. Experiment results show that the algorithm has good result.