Abstract: It has become critical to improve the recognition rate and security of the palmprint templates. In this paper, a cancelable palmprint template generating algorithm is proposed based on the secure sketch. Firstly, palmprint image feature is acquired using multi-resolution and multi-orientation Gabor filters to transform palmprint image. Its dimension is reduced and feature vector is extracted by the Principal Component Analysis（PCA） algorithm. Then a cancelable palmprint template is generated based on secure sketch, which fuses encrypted feature vector and BCH codeword. Both theoretical analyses and experimental results show that the method not only achieves good recognition performance, but also fulfils revocability, diversity and non-invertibility. Even the single secret-key is stolen or shared, the recognition rate remains high.

Key words: palmprint, cancelable template, secure sketch, Gabor filter, Principal Component Analysis（PCA）

摘要： 为了有效改进掌纹模板生成算法的识别率和安全性等性能，提出一种基于安全概略的可撤销掌纹模板生成算法。首先提取掌纹图像不同方向、不同尺度的Gabor幅值特征，并通过主成分分析（PCA）算法对其进行降维；然后将加密后的特征向量与BCH码异或融合，得到基于安全概略的可撤销掌纹模板。对比实验表明，该算法具有良好的识别性能，能较好地满足可撤销性、多样性和不可逆性，即便在单密钥丢失的情况下，也具有较高的识别率。

关键词: 掌纹, 可撤销模板, 安全概略, Gabor滤波, 主成分分析