Individual identification Based on Cascaded PCANet from ECG Signal
Jae-Neung Lee, Sung Bum Pan, Keun-Chang Kwak · 2019 International Conference on Electronics, Information, and Communication (ICEIC) · 2019
In this paper, we propose a Cascaded PCA Network (CPCANet) based on principal component analysis network (PCANet) for Individual identification of electrocardiogram (ECG). The CPCANet consists of three stages. In the first stage, ECG signals are preprocessed by normalization and spike removal. The R peak points in the preprocessed ECG signals are detected. Then, ECG signals are transformed into two-dimensional images to use as input of CPCANet. Here, we perform patch-mean removal and PCA algorithm as the same manner as PCANet from the transformed two-dimensional images. Also, we verified the influence of these parameters by modifying the key parameters of the experiment. The experiments are performed by the variation of patch size, number of filters, size of a block, and ratio of the lap from two databases. For this, we use a PTB ECG database which is benchmarking database. The experimental results clearly revealed that the proposed method showed good performance and effectiveness when compared with conventional algorithms such as PCA, auto-encoder (AE), extreme learning machine (ELM) and ensemble extreme learning machine (EELM).