一种基于特征捆绑计算模型的物体识别方法

Approach for Object Recognition Based on a Computational Model of Feature Binding

利用一种特征捆绑计算模型,以Gabor特征作为模型的初级特征,将相关统计量作为实现特征捆绑的基础,提出了一种物体识别方法.并实现了一组物体识别实验,结果显示,该方法能够进行较快速而准确地识别,说明了此方法和所使用的特征捆绑计算模型的有效性.

This paper proposes a novel method for object recognition by using a computational model of feature binding, in which Gabor features are employed as the elementary features and correlation statistics provide the basis for implementing the feature binding. A group of object recognition experiments are conducted with this method, and the results prove the comparatively good performances with high recognition precision and high speed, indicating the validity of this method and the computational model.