基于无线视觉传感器结合BP神经网络模型的枝干状态特征检测算法

Detection Algorithm of Branch State Feature Based onWireless Vision Sensor and BP Neural Network Model

由于检测木材枝干开裂、氧化等异常特征时存在检测误差高、运行时间较长的问题,文中提出基于无线视觉传感器结合BP神经网络模型检测枝干状态特征算法。利用无线视觉传感器结合BP神经网络模型并设计相关参数,以视觉传感器采集的木材枝干图像为样本,优化模型训练与参数。将枝干图像输入到BP神经网络模型中,利用模型计算分布曲线,通过曲线波峰波谷特征,判断各种异常特征,结合BP神经网络模型实现各种木材枝干状态异常数据检测。实验结果表明,在木材枝干异常数据检测时,数据可用度高达99%,检测精度高于85%,且运行时间较短。

Because of the problems of high detection error and long running time when detecting abnormal characteristics such as wood cracking and oxidation,this study proposes a detection algorithm of branch state feature based on wireless vision sensor and BP neural network model.Using wireless vision sensor combined with BP neural network model and designing the relevant parameters,taking the wood branch image collected by the visual sensor as the sample,the model training and parameters are optimized.The images of branches and stems are input into BP neural network model.The distribution curve is calculated by the model,and various abnormal features are judged by the peak and valley characteristics of the curve.Combined with the BP neural network model,the abnormal data detection of various wood branches and stems is realized.The experimental results show that the availability of data is up to 99%,the detection accuracy is higher than 85%,and the running time is short.