光学相干层析图像的散斑去躁和边缘特征提取

Noise removal and edge feature extraction of optical coherence tomography image

提出采用各向异性扩散(AD)算法对光学相干层析(OCT)图像作噪声消除和边缘增强的预处理,然后结合改进的相位一致(PC)算法进行边缘特征提取,并引入巴特沃斯高通提升滤波进一步衰减低频噪声,增强高频边缘细节.对珍珠和视网膜神经纤维层的OCT图像进行处理实验,结果表明,上述三种算法的综合应用,使OCT图像噪声得到有效抑制,轮廓清晰,特征部位得到加强。

The anisotropic diffusion （AD） algorithm was used to preprocess the optical coherence tomograpny （OCT） images for noise reduction and edge enhancement, then the phase congruency （PC） algorithm was added to improve the edge characteristics extraction, and finally Butterworth high-boost filter was appended for further edge details enhancing and noises clearing up. The novel combinations of the three algorithms were applied to process the OCT images from a pearl and retinal nerve fiber layer respectively. The experimental results showed that effective noise suppression, clear outline extraction, and characteristics enhancement can be realized.