基于MCA模型的指纹图像修复方法的研究

Research on Fingerprint Image Inpainting Method Based on MCA Model

在提取指纹图像的过程中,由于采集设备、环境等影响,往往采集到指纹图像不仅含有噪声而且有不同程度的纹线断裂,现有的指纹图像处理方法大多只做了去噪处理,没有对指纹图像进行修复,导致得到的指纹图像含有的特征点较少,降低了指纹识别的正确率。为解决这一问题,论文根据指纹图像的特点,在经典的偏微分方程修复模型和Starck等提出的MCA图像修复模型的基础上进行改进,采用了CDD模型和MCA模型相融合的算法,不仅解决了经典偏微分方程图像修复模型对纹理区域修复产生的虚假的现象,又克服了原MCA修复模型无法满足连通性的缺点,适合指纹这种复杂纹理的图像修复,实验结果证明,该算法相对于传统修复算法在指纹断裂痕的修复上具有更好的效果。

In the process of fingerprint image extraction,due to the influence of acquisition equipment and environment,the fingerprint image not only contains noise,but also has different degrees of ridge fracture.Most of the existing fingerprint image pro⁃cessing methods only do denoising,and do not repair the fingerprint image,resulting in less feature points in the obtained finger⁃print image,which reduces the accuracy of fingerprint recognition.In order to solve this problem,according to the characteristics of fingerprint image,this paper improves the traditional partial differential equation(PDE)model and the MCA image inpainting mod⁃el is proposed by Starck and so on,and adopts the fusion algorithm of CDD model and MCA model,which not only solves the false phenomenon of the classical PDE image inpainting model for texture region,but also overcomes the original MCA repair model.The experimental results show that this algorithm is better than the traditional repair algorithm in fingerprint crack repair.