| Title |
The 1D Wada index for the classification of digital images of concrete cracks / |
| Authors |
Orinaitė, Ugnė ; Ragulskienė, Jūratė |
| DOI |
10.15388/DAMSS.13.2022 |
| ISBN |
9786090707944 |
| eISBN |
9786090707951 |
| Full Text |
|
| Is Part of |
DAMSS 2022: 13th conference on data analysis methods for software systems, Druskininkai, Lithuania, December 1–3, 2022 / Lithuanian computer society, Vilnius university Institute of data science and digital technologies, Lithuanian academy of sciences.. Vilnius : Vilnius university press, 2022. p. 71.. ISBN 9786090707944. eISBN 9786090707951 |
| Abstract [eng] |
The Wada index has been recently introduced for the detection if a given basin boundary is a Wada boundary. The Wada index is based on the weighted and truncated Shannon entropy and does represent the number and the distribution of different colours (attractors) in the twodimensional phase space of initial conditions. The Wada index is based on the standard box counting algorithm. That makes the algorithm for the computation of the Wada index conveniently applicable for different basins of attraction represented as color digital images. With the recent advances in machine learning, the development of ANN- and CNN-based algorithms has become a popular approach for the automated detection and identification of concrete cracks. However, most of the proposed models are trained on images taken in ideal conditions and are only capable of achieving high accuracy when applied to the images of concrete cracks devoid of irregular illumination conditions, shadows, shading, blemishes, etc.. A 1D modification of the Wada index is presented in this paper. It is demonstrated that the 1D Wada index algorithm can be used as an efficient pre-processing tool for digital images contaminated by the additive and/or optical noise. The 1D Wada index algorithm helps to reduce the additive noise in images of concrete cracks what enables better classification based on deep learning algorithms. Alexnet convolutional neural network (CNN) is used to train the classification model on Mendeley Concrete Crack Images for Classification dataset. It is demonstrated that the application of the 1D Wada index algorithm helps to improve the classification accuracy of concrete crack images contaminated by noise up to 98% what corresponds to the industrial standard in the field of autiomatic crack identification. |
| Published |
Vilnius : Vilnius university press, 2022 |
| Type |
Conference paper |
| Language |
English |
| Publication date |
2022 |
| CC license |
|
