基于皮带出渣图像识别渣土含水率区间

Identifying the water content interval of muck based on the image of belt slag

为实时识别渣土含水率,通过制备3种初始含水率细砂,添加不同泡沫注入比的泡沫制成不同含水率的改良渣土,通过皮带出渣试验平台开展出渣试验,获取皮带上渣土图像,并采集渣土样测定其含水率,以1%为间隔标记含水率区间,建立渣土图像与含水率区间数据集。通过图像预处理,采用简化局部像素强度模式结合完备局部二值模式的方法提取渣土主体图像与边缘图像纹理特征,选取粒子群优化的支持向量机模型作为基模型,进一步构建渣土含水率识别集成学习模型,提高了识别准确率,含水率识别误差为±1%。

In order to identify the soil water content in real time,the improved muck with three kinds of fine sand with initial water content were prepared by adding foam with different foam injection ratios,the slag experiment was carried out through the belt slag test platform,the muck images on the belt were obtained,the muck samples were collected accordingly to determine the water content,the water content interval was marked at 1%intervals,and the data set of muck images and water content intervals was established.Through image preprocessing,the texture features of the main image and the edge image of the muck were extracted by using the method of simplified local intensity order pattern combined with completed local binary pattern,and the support vector machine model of particle swarm optimization was selected as the base model,and the integrated learning model for the recognition of water content of the muck was further constructed,which improved the recognition accuracy,and the recognition error of the water content was±1%.