Search

The machine vision system was employed to inspect the surface defects of bridge cables of cable-stayed bridges. After the acquisition and preprocessing of the defect images, it is necessary to classify and identify the defects of the cables to meet the requirements of non-destructive testing and evaluation. In this paper, feature extraction for defect images was performed using mathematical statistical methods. After that, 10 feature parameters including shape features, grayscale features and texture features of the defect images were obtained and selected for a classification model of support vector machine (SVM). To improve the SVM classification performance, the particle swarm optimization algorithm (PSO) was adopted to obtain the punish factor c and the kernel parameter g of the SVM model, namely the PSO-SVM algorithm. Finally, our PSO-SVM classification model was employed to implement the classification of real surface defect images of the bridge cables. Longitudinal crack, transverse crack, surface corrosion, and pothole defect can be automatically identified and the classification accuracy reached 96.25%. The experimental results showed that the PSO-SVM model can improve the classification performance of the surface defects. Based on the effective classification, we can find the distribution characteristics of the surface defects of the cable. It is very important to analyze the relationship between the type of surface defects and the material of the protective layer, so as to adopt appropriate materials and reasonable maintenance measures. Thus, it has a great significance in the structural health monitoring of bridge cables.

Deep neural networks have received increasing attention in human action recognition. Previous research has established that utilizing 3D convolution is a reasonable approach to learn spatio-temporal representation. Nevertheless, constructing effective 3D ConvNets usually need an expensive pre-training process that performing on a huge-scale video dataset. To avoid this burdensome situation, one major issue is to determine whether the pre-trained parameters of 2D convolution networks can be directly bootstrapped into 3D. In this paper, we devise a 2D-Inflated operation and a parallel 3D ConvNet architecture to solve this problem. The 2D-Inflated operation is used for converting pre-trained 2D ConvNets into 3D ConvNets, which avoiding video data pre-training. We further explore the optimal quantity of 3D ConvNet in the parallel architecture, and the results suggest that 6-nets architecture is an excellent solution for recognition. Another contribution of our study is two practical and valid skills, accumulated gradient descent and video sequence decomposition. Either of those techniques can promote the improvement of performance. The recognition results of UCF101 and HMDB51 reveal that, without the video data pre-training, our 3D ConvNets still can achieve competitive performance to the other generic and recent methods of using 3D ConvNets in the RGB image domain.