基于Canny优化算法物理模拟实验覆岩裂隙场自动识别研究

为了实现实验过程中采动覆岩裂隙信息的自动识别，以某矿15235工作面为工程背景，设计了相似材料模拟实验，通过对定焦相机批量获取的近景摄影测量影像进行拼接和校正，获得了研究区域的正射影像。针对Canny算法在影像信息阈值分割方面的不足，提出了一种自适应阈值分割优化算法，实现了相似材料模型表观裂隙场的自动识别。基于Matlab研发平台，分别采用像素法和计盒维数法对工作面不同推进度的覆岩裂隙场进行了定量分析。结果表明，裂隙场信息提取完整，当工作面推进度分别为46、118、255 cm时，类间方差权重系数为3.16％、5.73％、7.68％，覆岩裂隙场分形维数为1.804、1.811和1.885，采动覆岩裂纹生长具有良好的自相似分形特性。工作面推进至255 cm时，采动覆岩垮落带高度和断裂带高度分别为13.31 cm和63.36 cm，左右垮落角分别为52°和55°。采用改进的Canny算法能够实现物理模拟实验采动覆岩裂隙场的精确辨识。

Research on automatic recognition of overburden fracture field based on Canny optimization algorithm in physical simulation experiment

In the interest of auto-identification of overburden strata fracture in the process of analogue test，sand-plaster physical model was developed against the backdrop of 15235 working coal panel.Batch images obtained by close-range photogrammetry with camera focus fixed were montaged and calibrated.Thereafter digital orthophoto maps of the whole physical simulation model were acquired.Aiming at the deficiency of Canny algorithm on the aspect of image information threshold segmentation，an self-adaptive threshold segmentation optimized- algorithm is proposed to detect the fracture field of sand-plaster model surface.Based on the platform Matlab，the methodology of pixel and box-counting dimension were adopted to quantitatively analyze overlying strata fracture field induced by coal mining for different mining sizes.The results show that the fracture information is integral with mining sizes being 46 cm，118 cm and 255 cm respectively，the weight coefficients of the inter class variance are 3.16％、5.73％ and 7.68％，with fractal dimension being 1804、1811 and 1885 respectively.The mining-induced fractures have favorable self-similar fractal features.When mined-out face size is 255 cm，the heights for caved zone and fractured zone of overburden strata are 13.31 cm and 63.36 cm with left and right angles of draw being 52°and 55° respectively.The improved Canny algorithm can precisely identify mining-induced fractures on the surface of physical model.