基于机器视觉的长壁工作面直线度测量算法研究

Straightness measurement algorithm based on machine vision for coal longwall face

综采工作面推进过程中,由于上百米长的狭长地形和起伏弯曲,其直线度是一种大型移动目标的非可视测量问题。通过将梯形窗口、灰度识别和特征搜索等视觉算法应用于实时测量刮板输送机的形态,建立视觉相机测量直线度的局部坐标系,可使测量的结构光点像素值转换为设定直线度的偏差距离;通过链位传递算法将工作面布置的多个视觉相机的局部直线度坐标经过转换矩阵和平移向量建立全局坐标系,从而计算出以刮板机节为测量标志点的工作面直线度。实验表明,视觉检测整个工作面直线度达到±50 mm的精度,满足采煤机沿平直工作面高效运行的要求。

In the mining process of fully mechanized coal face with few hundred meters,long,narrow and undulating curved,face straightness is a non-visual measurement of large moving target. Studying the trapezoidal windows,grayscale,recognition and feature search vision algorithms as real-time measuring form of scraper conveyor,the straightness of local coordinate system is established with measurement of structure light pixel value to a straightness deviation from set. By chain-transfer algorithm,the face layout of multiple local straightness with camera coordinate system is transformed to global coordinate system with matrices and translation vector,as the conveyor section is the face of straightness measurement markers. Experiments show that visual inspection achieves accuracy of ± 50 mm for the whole face straightness,it meets requirements of highly effective shearer running along straight face.