纹理边界引导的复合材料圆孔检测方法

针对复合材料纹理背景下圆形基准孔检测困难的问题，提出一种纹理边界引导的圆孔检测方法.该方法将局部三进制模式（LTP）与灰度共生矩阵（GLCM）对比度融合为纹理对比度，通过提取纹理对比度特征实现孔内纹理和孔外纹理的快速分割，得到与圆孔边界近似吻合的纹理边界，利用纹理边界的位置信息去除绝大多数非圆孔边界的边缘点，利用纹理边界的连通信息对剩余边缘点进行分组，使用内嵌圆参数统计机制的随机圆检测算法从每组边缘点各检出1个圆孔目标，进而完成对多个圆孔目标的检测.实验结果表明，在复合材料圆孔检测场景中该方法有94%以上检出率，3%以下检错率和较高的检测速度，并表现出良好的检测鲁棒性.

Detection method for composite hole guided by texture boundary

A novel hole detection method guided by texture boundary was proposed, aiming to solve the detection problem of composite circular reference holes. The texture contrast based on local ternary pattern (LTP) and gray-level co-occurrence matrix (GLCM) was extracted first to perform the fast segmentation of intra-hole texture and out-of-hole texture, which could obtain the texture boundaries closely matching the hole boundaries. Then, the position information of texture boundaries was used to remove most of the edge points that belong to non-hole boundaries. The connectivity information of texture boundaries was used to group the remaining edge points. Finally, the randomized circle detection method with circle-parameter statistics mechanism was applied to detect a single circle from each group of edge points, which completed the detection of multiple hole targets. Experimental results showed that this method had a detection rate of more than 94%, an error rate of less than 3%, a high detection speed and good detection robustness in the composite hole detection scene.