To ensure traffic safety, pavement conditions should be evaluated and distress (cracks, potholes, etc.) should be timely detected. In recent years, methods based on the analysis of digital images have been proposed to automatically detect pavement distress. However, most of these methods process the images offline and therefore require a large amount of data to be stored until actual processing. To enable real-time analysis of the images and to reduce the amount of stored data, a highly performant and computationally inexpensive implementation of a current analysis method is required. This paper presents a Graphics Processing Unit (GPU) implementation of an image analysis method for pavement distress detection. GPUs have been recently utilized for high-performance computing in diverse scientific fields. The pavement distress detection method presented in this paper is based on the wavelet transform. The implementation is carried out using the Open Computing Language (OpenCL). To evaluate the performance, the method was tested on 30 pavement images. The results show that a significant improvement in performance can be achieved by utilizing GPUs.
[1]
Nagavijayalakshmi Vydyanathan,et al.
Parallel discrete wavelet transform using the Open Computing Language: a performance and portability study
,
2010,
2010 IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum (IPDPSW).
[2]
Jian Zhou,et al.
Wavelet-based pavement distress detection and evaluation
,
2003
.
[3]
Ulrich Rüde,et al.
Fast Wavelet Transform Utilizing a Multicore-Aware Framework
,
2010,
PARA.
[4]
A. Haar.
Zur Theorie der orthogonalen Funktionensysteme
,
1910
.
[5]
Ingrid Daubechies,et al.
The wavelet transform, time-frequency localization and signal analysis
,
1990,
IEEE Trans. Inf. Theory.