利用稀疏化生物视觉特征的多类多视角目标检测方法

鉴于生物视觉特征在目标分类中的良好性能，采用了一种基于稀疏化生物视觉特征的多类多视角目标检测方法。首先利用特征稀疏化方法对生物视觉的标准模型进行改进，有效提高了目标的可分性，然后利用滑动窗口的方法构建基于稀疏化生物视觉特征的目标检测器，采用局部邻域抑制算法完成特定目标的检测任务，最后通过构建场景中待检测目标的词典，对每类目标分别设计滑动检测器以完成多类多视角目标的检测任务，实验结果表明，该方法具有很好的检测性能。

Multiclass and multiview object detection approach based on sparse biological vision features

As the biological vision features shows superior performance on object classification,a multiclass and multiview object detection approach based on sparse biological vision features was adopted.Firstly,the standard model of biological vision was improved with the technique of sparse features,which improved the separability of object effectively.Then,the object detector based on sparse biological vision features was designed with the technique of sliding window,and the detection task was completed via local neighborhood suppression algorithm.At last,the multiclass and multiview object detection task was accomplished through building object dictionary and designing several object detectors in the scene.The experimental results show that the proposed approach exhibits a robust performance.