Highway Roadbed Void Detection Algorithm Based On Wavelet Analysis

Ground penetrating radar, used in road non-destructive detection, is a new technology developed in recent years. Because of its high measuring precision, it can meet the demand of measuring error, and has been widely used at home and abroad. However the analytical method of GPR's reflected signals mostly still remain in the manual detection stage. This paper attempts to apply the wavelet transformation to processing and analyzing the reflected signals of GPR, and realize automatic detection of cavities in the roadbed.This paper firstly analyzes the reflected signals of GPR, and summarizes the relations between singular points in reflected signals and cavity positions. Then according to some singular point's characteristics in modulus maximum by wavelet transformation, the paper proposes a method of detecting the singular points in reflected signals, and proves the rationality of this method. Next, the paper uses the wavelet transformation modulus maximum value, the modulus maximum value curve as well as the signal lipschitz-index to detect singular points in reflected signals and realizes the algorithm of automatic detecting cavities in roadbed. The algorithm mainly consists of two steps: first, pre-processing GPR signals, which mainly transform data and remove noise component in the signal by median filter; Second, according to characteristics of singular points in frequency area, using modulus maximum value of wavelet transform to find out singular points in reflected signals, and then realize automatic detection of cavities in roadbed by singular points and similitude degree. Otherwise, based on the characteristics of reflected signals of road, we use trace pertinence to improve continuity of nearby data, in order to remove disturbance and improve veracity.Proved by practice, the algorithm which this paper studies has some effect, but need further research on the method of estimating cavity position by singular points, and in this foundation's improvement will be the key point which our next research focus on.