基于模板匹配和深度学习的反光弧面缺陷检测

Reflective arc surface defect detection based on template matching and deep learning

针对反光弧面表面具有很强的镜面反射特性,在对此类工件表面检测时非常困难这一难题,提出了基于模板匹配和深度学习的检测方法。在平滑漫射光照条件下,通过不同角度获得带有条带纹理特征,然后经过图像处理获得其轮廓,与标准的特征轮廓模板进行匹配识别。还可以将获得的图片制成数据集,训练一个深度学习模型来对获得图片进行缺陷识别。实验结果表明:基于模板匹配和深度学习的检测方法均能有效对缺陷进行识别。

Aimiong at the problem that it is very difficult to detect the defect of reflective arc surface due to the surface of reflective arc surface has strong reflective characteristics,a detection method based on template matching and deep learning is proposed.Under the condition of smooth diffuse illumination,stripe texture features are obtained from different angles,the contour is obtained by image processing.It is matched with the standard feature contour template and recognition.The obtained images can also be made into datasets,and a deep learning model is trained to identify the defects of the obtained images.Experimental result show that the detection methods based on template matching and deep learning can effectively identify defects.