Crack Detection and Geometry Measurements using Digital Image Processing

Abstract

Digital image processing (DIP), is the use of a digital computer to process digital images through an algorithm. In this work DIP used to perform crack/defect Detection and geometry measurements in terms of angle, length, and width by building a MATLAB code. In this research, the test of tensile were applied on two flat samples, the two flat samples are painted in a speckle pattern on the tested surface to use DIP features. The DIP gives a low-cost optical, and an appropriate technique to detect crack/defect as well as geometry measurements with the load increment in terms of angle, length and width without any contact with the tested surfaces. The tests directed for both of flat Aluminum alloy and Copper alloy specimens. In order to guarantee the efficiency of the planned DIP systems, the comparison is considered between the test results of this technique with the measurements gained by actual physical by using calibrated digital vernier calliper (IP 54) to measure the length of the crack with 0-300 mm range and 0.01 mm resolution. The DIP obtained percentage of accuracy varies from (97) % to (99) % by comparison with the measurements gained by actual physical way.

Leveraging cross-view geo-localization with ensemble learning and temporal awareness

“…Then the trajectories that have blurry aerial tiles (13.5% of the bright trajectories) are removed (similar to the example shown in Fig 4 ). A Laplacian filter [ 53 ] with a threshold of 200 is used to detect blurry aerial tiles, and a gray scale mean filter with a threshold of 85 is employed to detect dark ground frames. Both thresholds are chosen empirically to drop all the true positive corrupt examples with some false positives.…”