近邻系数协同强化人脸图像子空间聚类法

针对最小二乘回归子空间聚类法没有考虑近邻样本对求解表示系数的影响这一不足,提出近邻系数协同强化子空间聚类法.该方法利用近邻样本相似导致表示系数接近的思想定义近邻系数协同强化项.通过近邻样本的系数强化表示系数,从而得到更能反映样本相似度的相似矩阵,进而提高聚类准确率.在6个人脸图像数据集上的实验表明该方法是有效的.

Nearest neighbor coefficient cooperative reinforcement subspace clustering method for face image

In order to solve the problem that the least square regression subspace clustering method does not consider the influence of the nearest neighbor samples on solving the representation coefficient, the nearest neighbor coefficient cooperative reinforcement subspace clustering method is proposed. The proposed method uses the idea that the similarity of neighbor samples leads to the similarity of representation coefficients to define the cooperative reinforcement term of neighbor coefficients. The coefficients are reinforced by the coefficients of the nearest neighbor samples, so as to obtain the similarity matrix which can better reflect the similarity of samples, and then improve the clustering accuracy. Experiments on six face image datasets show that this method is effective.