增强现实诱导维修进程识别策略

针对传统的增强现实维修系统不能有效对维修状态进行感知和判断的问题进行了研究，提出了一种基于概率神经网络（probabilistic neural network, PNN）的维修进程识别的策略。该策略将关注的焦点由维修对象本身转移到附近的已拆卸零件放置区域，利用曲波变换生成已拆卸零件边缘图像，然后生成Hu不变矩特征值，再将特征值作为PNN神经网络的输入进行进程零件的分类识别，从而得到当前的维修进程。实验表明，该策略下识别的准确度明显高于图像匹配方法与曲波不变矩方法，为场景感知提供了新的思路和解决方案。

Recognition strategy for augmented reality induced maintenance process

In view of the problem that the traditional augmented reality maintenance system can not effectively perceive and judge the state of maintenance, this paper proposed a method based on PNN(probabilistic neural network) for maintenance process identification. The strategy shifts the focus from the maintenance object itself to the nearby removed parts placement area, uses the curvelet transform to generate the edge image of the disassembled parts, then generates the Hu invariant moment characteristic value, which is used as the input of PNN (probabilistic neural network) neural network to classify and identify the parts of the process, finally gets the current maintenance process. Experimental results show that the recognition accuracy of the proposed method is significantly higher than that of the image matching method and curvelet invariant moment method, which provides a new idea and solution for the scene perception.