The Classification of Pavement Crack Image Based on Beamlet Algorithm
Pavement distress, the various defects such as holes and cracks, represent a significant engineering and economic concern. This paper based on Beamlet algorithm using MATLAB software to process the pavement crack images and classify the different cracks i
- PDF / 1,055,863 Bytes
- 9 Pages / 439.363 x 666.131 pts Page_size
- 99 Downloads / 210 Views
Abstract. Pavement distress, the various defects such as holes and cracks, represent a significant engineering and economic concern. This paper based on Beamlet algorithm using MATLAB software to process the pavement crack images and classify the different cracks into four types: horizontal, vertical, alligator, and block types. Experiment results show that the proposed method can effectively detect and classify of the pavement cracks with a high success rate, in which transverse crack and longitudinal crack detection rate reach to 100%, and alligator crack and block crack reach more than 85%. Keywords: Pavement Crack, Beamlet Algorithm, Classification, Transform.
1
Introduction
Pavement distress, the various defects such as cracks illustrated in Fig.1, represent a significant engineering and economic concern. Pavement crack image classification is important in an automated pavement inspection system, because it can provide critical information for pavement maintenance. There has been a significant amount of research during the last two decades in developing image processing algorithm for pavement crack inspection. Chou et al [1]. approached the problem of pavement crack classification by using moment invariant and neural networks. After preprocessing and thresholding into binary images, they calculated Hu, Bamieh, and Zemike moments. Teomete et al [2]. proposed histogram projection to identify cracks within a cropped image. While focused on the severity of cracks, crack classification, was not performed. Moreover, the system cannot detect multiple cracks within an image.In a paper by Bray [3], cracks is performed using a neural network while classification is performed by another neural network. The proposed algorithm has not been tested on real images. Cheng et al [4]. described a neural network based thresholding method to segment and classify pavement images that can be implemented in real time. Tsai et al [5]. presented a critical assessment of various segmentation algorithms for pavement distress detection and classification. *
Corresponding author.
D. Li and Y. Chen (Eds.): CCTA 2013, Part II, IFIP AICT 420, pp. 129–137, 2014. © IFIP International Federation for Information Processing 2014
130
A. Ouyang et al.
Alligator
Block
Longitudinal
Transverse
Fig. 1. Types of pavement cracks
2
Beamlet Alogrithm
2.1
Beamlet Dictionary
The beamlet transform is performed in the dynamically partitioned squares of an image. Images are viewed as the continuum square [0, 1]2 and the pixels as an array of 1/n by 1/n squares arranged in a grid in [0, 1]2. The collection of beamlets is a multiscale collection of line segments occurring at a full range of orientations, positions, and scales[6], as illustrated in Fig.2. 2.2
Beamlet Transform
The beamlet transform is defined as the collection of line integrals along the set of all beamlets. Let f ( x1 , x2 ) be a continuous function on 2-D space, where x1 and x 2 are coordinates. The beamlet transform
T f of function f is defined as follows:
T f (b) = f ( x
Data Loading...
