基于改进U-net网络的液压管路分割方法

针对液压管路背景多变、管道弯折、管道重叠排布等复杂现象且现有图像分割方法对管路分割精度不高等问题,提出一种以U-net网络为基础,结合Mobilenetv3网络、SE注意力机制模块、自校正卷积模块的液压管路分割方法。该方法以Mobilenetv3-large模型作为骨干网络,结合LR-ASPP网络处理特征图;在解码过程中,融入SE注意力模块和SC自校正模块,提升了特征提取能力;最后采用Dice函数和BCE函数的组合来作为网络的损失函数,有效地提升了网络的收敛能力。实验结果表明本文提出的方法在交并比、像素精度指标上的均值分别达到90.8%、95.2%,且模型体积为16.9 M,推理每张图像所耗时间20 ms,可应用于需实时部署的场景,为液压管路渗漏的准确识别提供了基础。

Hydraulic pipeline segmentation method based on improved U-net network

Aiming at the complex phenomena such as variable background, bending and overlapping arrangement of hydraulic pipelineand the low accuracy of pipeline segmentation by existing image segmentation methods, a hydraulic pipeline segmentation method basedon U-net network, combined with Mobilenetv3 network, squeeze-and-excitation networks module and self-calibration convolutionalmodule is proposed. The method selected Mobilenetv3-large model as the backbone network, and the feature maps are processed withLraspp network. In the decoding process, the squeeze-and-excitation networks module and self-correction module are integrated toimprove the feature extraction ability. Finally, the combination of Dice function and BCE function is used as the loss function of thenetwork, which effectively improves the convergence ability of the network. Experimental results show that the mean values of theproposed method in the intersection over union and pixel accuracy are 90. 8% and 95. 2%, respectively. The model size is 16. 9 M, andthe reasoning time for each image is 20 ms, which can be applied to the scene requiring real-time deployment. It provides a basis for theaccurate identification of hydraulic pipeline leakage.