Micro crack detection with Dijkstra’s shortest path algorithm

Abstract

A package based on the free software R is presented which allows the automatic detection of micro cracks and corresponding statistical analysis of crack quantities. It uses a shortest path algorithm to detect micro cracks in situations where the cracks are surrounded by plastic deformations and where a discrimination between cracks and plastic deformations is difficult. In a first step, crack clusters are detected as connected components of pixels with values below a given threshold value. Then the crack paths are determined by Dijkstra's algorithm as longest shortest paths through the darkest parts of the crack clusters. Linear parts of kinked paths can be identified with this. The new method was applied to over 2,000 images. Some statistical applications and a comparison with another free image tool are given.

Statistical analysis of damage evolution with a new image tool

“…5). More details of the method and the output are given in Gunkel et al 26 . But not all detected crackclusters and crackpaths are related to real cracks.…”

“…Gunkel et al 26 . proposed a new crack recognition method which follows the darkest parts of a damage area.…”

“…For a detailed crack analysis, it is also important to know what the kinks and curves of a crack are. Gunkel et al 26 proposed a new crack recognition method which follows the darkest parts of a damage area. Although it is based on the damage areas, it has the ability to provide all kinks and curves of a crack.…”

An Improved Neuron Segmentation Model for Crack Detection – Image Segmentation Model

“…Among them, the minimal path-based methods [16,[21][22][23][24][25], which appear to be most prevalent in recent literature, consider a crack as a path of minimal length in a certain metric and then convert the crack detection problem to that of finding the minimal path. Similarly, the cracks with complicated geometrical changes such as branches and curves could also be coped in [26][27][28].…”

Multipath Routing Algorithm for Video Stream Transmission Based on Clustering Theory

“…SPEED routing protocol is a real-time routing protocol, through the link delay estimation mechanism, to predict the network congestion, and make use of the local geographic information of the node and the transmission rate to make the routing decisions. MMSPEED adds the multipath transmission mechanism on the basis of the SPEED, so as to improve the reliability of the routing algorithm (Gunkel, Stepper, Müller and Müller, 2012;Sabri, Basari, Husin and Samah, 2014). Multipath routing ReInForM method ensures the reliability of data transmission through its more copies by sending multiple copies of the data packet.…”