摘要： 多类物体识别在提取特征之后，样本的数量会呈指数倍增加，为减少计算量同时，不降低识别率，采用亲和传递算法对样本数据进行聚类形成视觉字典，帮助并提升物体识别效率。在Sowerby图像数据库上进行实验证明，该方法与使用k均值聚类建立视觉字典方法相比，在同等条件下具有更高的识别率。

关键词: 亲和传递聚类, 多类物体识别, 视觉字典

Abstract: The multi-class object recognition method usually suffers from the problem of curse dimensionality after extracting feature. This paper proposes a new approach for learning a discriminative model of object classes combined affinity propagation clustering. By using Affinity Propagation(AP) , a representative visual vocabulary can be obtained. Experiment in Sowerby databases shows that it is superior to the recognition rate of k-means algorithms .

Key words: Affinity Propagation(AP) clustering, multi-class object recognition, visual vocabulary

中图分类号: 

• TP18