基于HOG—AP的人脸图像识别算法

针对仿射传播（AP）算法存在缺乏判定最优聚类结果的指标以及收敛性能不够好等问题，提出了一种基于方向梯度直方图（HOG）的AP改进算法。首先提取图像的HOG特征向量，然后引入收缩因子加速仿射传播算法的收敛过程，最后将有效性指标嵌入算法的迭代过程，监督并引导算法向着最好聚类质量的方向运行。对人脸图像进行实验，实验结果表明，基于HOG的AP改进算法可以得到更接近正确类数的结果，提高了FM值，降低了错误率。

Human face image recognition algorithm based on HOG-AP

Affinity propagation （AP） algorithm has the problems about there is no a index to know what is the optimal solution and the convergence performance of affinity propagation algorithm is bad. Therefore,this paper proposes an improved algorithm of affinity propagation algorithm based on histograms of oriented gradients. Firstly,this paper extrats histograms of oriented gradients of images,then it introduces constriction factor to accelerate convergence of affinity propagation algorithm,finally it embeds validity index in iterative process of algorithm to supervise and guide algorithm to run in the direction for the best clustering quality.Experiments are conducted on human face images,experimental results show that improved affinity propagation algorithm based on histograms of oriented gradients can get more accurate clustering number than affinity propagation algorithm, Fowlkes Mallows value is improved, and the error rate reduces.