基于Faster R-CNN的隧道图像裂缝检测

Crack of Tunnel Detection Based on Faster R-CNN

近几年,随着城市隧道建设力度加大,隧道已进入中年期,其健康问题不容忽视。目前,基于图像处理技术的隧道裂缝检测已成为一种便捷的检测手段。然而,由于隧道图像存在对比度低、光照不均匀、噪声多等问题,对传统图像处理方法造成较多挑战,对此提出了一种基于深度学习网络Faster R-CNN的隧道图像裂缝检测方法。首先利用大量隧道裂缝图像训练产生裂缝模型,然后用此模型对隧道图像进行裂缝检测。实验结果表明,该方法能够弱化隧道图像质量问题,实现了对隧道图像裂缝的快速检测和准确定位与标注。

In recent years, with the increase of the construction of urban tunnels, the tunnel has entered the middle age, and its health problems cannot be ignored. At present, the crack of tunnel detection based on image processing technology has become a convenient detection method. However, due to the problems of low contrast, uneven illumination, and high noise in the tunnel images, the traditional image processing methods face more challenges. This paper proposes a tunnel crack detection method based on Faster R-CNN. Firstly, a large number of tunnel crack images are used to train the model, and the model is used to detect cracks in the tunnel image. The experiments show that this method weakened the image quality problem, and the fast detection and accurate positioning and labeling of tunnel image cracks can be realized.