基于光电传感器的目标轮廓提取及其识别

Object Profile Extraction and Recognition Based on Photoelectric Sensors

针对物体外形轮廓特征,给出一种基于主动式光电传感器的轮廓特征提取方法,该方法是通过物体经过传感器组视场时各传感器输出状态来获取特征信息;并结合稀疏表示理论,提出一种基于稀疏表示的目标轮廓识别算法,该算法是将测试样本投影到稀疏域,依据系数的稀疏性对测试样本进行分类。数值仿真与实验结果表明:采用光电传感器组获取的信息能够有效地描述物体;且通过本文算法对不同类型物体进行分类,具有良好的识别效果;对不同条件下获得的特征样本进行分类时,具有良好的鲁棒性。

According to profile features of an object,a kind of object profile extraction system composed of active photoelectric sensors was proposed, which acquired the feature information by output state of sensors when objects passed through sensors field of view, and an algorithm of profile recognition which combined the theory of sparse representation was proposed. As the first step, the algorithm projected the testing samples into sparse domain, which recognized the samples according to coefficient sparsity. Numerical simulation and experiments showed the information obtained from photoelectric sensor group can effectively describe objects, and the algorithm had good classification effect and robustness under different conditions.