Development of MATLAB GUI Based on Blob Analysis for Automatic Extraction of Defects

This study introduces a graphical user interface (GUI) based on MATLAB to realize the automatic ex-traction of sizes of defects from the infrared sequence. To obtain the edge of the defect at deeper layer, a fusion stratagem of the maximum of gray values is adopted for an image subset in the sequence. Blob analysis to the fusion image is used to obtain the general information of defects of a specimen including the distributions and numbers of defects. The frame image is determined for a certain defect according to the peak of the time history curve of sensitive region variance. It can yield a region of interest (ROI) to expand the blob in the selected frame and the defect can be acquired by image segmentation. The results show that through this GUI, a better thermal image can be selected from a set of infrared sequence diagrams for quantitative extraction of different buried depth defect areas, which realizes automatic defect extraction, and its relative error is within 5%. The research on infrared automatic detection technology has certain significance.

Improved Pixel-Level Pavement-Defect Segmentation Using a Deep Autoencoder

Convolutional neural networks perform impressively in complicated computer-vision image-segmentation tasks. Vision-based systems surpass humans in speed and accuracy in quality inspection tasks. Moreover, the maintenance of big infrastructures, such as roads, bridges, or buildings, is tedious and time-demanding work. In this research, we addressed pavement-quality evaluation by pixelwise defect segmentation using a U-Net deep autoencoder. Additionally, to the original neural network architecture, we utilized residual connections, atrous spatial pyramid pooling with parallel and “Waterfall” connections, and attention gates to perform better defect extraction. The proposed neural network configurations showed a segmentation performance improvement over U-Net with no significant computational overhead. Statistical and visual performance evaluation was taken into consideration for the model comparison. Experiments were conducted on CrackForest, Crack500, GAPs384, and mixed datasets.