一种煤矿井下输煤大块物检测方法

针对现有煤矿井下输煤大块物检测方法存在无法检测大块物数量且检测精度不高的问题,提出了一种基于改进HED神经网络融合Canny算子的煤矿井下输煤大块物检测方法。采用提取反射分量结合边缘保留滤波方法对采集的图像进行预处理,增强图像亮度、对比度,加深图像边缘信息;将预处理的图像代入改进的HED神经网络与Canny算子的融合模型中,得到连续的大块物边缘图像,根据边缘图像做非运算得到二值化填充图像;对二值化填充图像中的大块物进行矩形标注,计算出大块物像素个数与面积;统计大块物数量并判断大块物面积是否高于设定阈值,若高于设定阈值,则报警。实验结果表明,基于改进HED神经网络融合Canny算子的煤矿井下输煤大块物检测方法具有较好的边缘检测效果,能够有效降低图像边缘检测误差,有效统计出大块物数量,并能计算出大块物的面积。

A detection method for large blocks in underground coal transportation

In view of problems that the existing detection methods for large blocks in underground coal transportation cannot detect quantity of large blocks and detection accuracy is not high,a detection method for large blocks in underground coal transportation based on improved HED neural network and Canny operator was proposed.Firstly,extracted reflection component combined with edge reservation filtering method is used to preprocess collected image,so as to enhance the image brightness and contrast and deepen image edge information.Then,the preprocessed image is substituted into the fusion model of improved HED neural network and Canny operator to obtain the continuous large blocks edge image,and the binarization filled image is obtained by doing non-operation according to the edge image.The large blocks in the binarization image are marked with rectangle and the pixel number and area of the large blocks are calculated.Finally,the number of large blocks is counted and judge whether the area of large blocks is higher than the set threshold.If the area is higher than the set threshold,the alarm will be given.The experimental results show that the detection method for large blocks in underground coal transportation based on improved HED neural network with Canny operator has a good edge detection effect,which can effectively reduce the image edge detection error,and count the number of large blocks,calculate the area of large blocks.