The groove is the biggest interference factor in the identification of cement pavement cracks.In the process of restraining or eliminating the grooves,it inevitably reduces the contrast between the cracks and the background,at the same time weakens the cracks in the grooves and destroys the connectivity of the cracks.A large number of short crack fragments are mixed with noise,which increases the difficulty of subsequent identification.In order to solve these problems,a slot elimination method based on DFT transform and a splicing method based on grid voting method are proposed.Firstly,the image of cement pavement is transformed from DFT to frequency domain,and the characteristic of groove in frequency domain spectrum is obtained.On this basis,according to the periodicity of the groove in the spatial image,the position of the groove spectral peak in the frequency domain image is deduced,and a series of filters are designed to eliminate these spectral peaks in the frequency domain,and then the slot is eliminated by the inverse DFT transform in the spatial domain.Then,the global spatial image of cement pavement is binarized by Otsu method,and the noise is filtered after the square vote to realize the splicing of cracks so as to achieve the purpose of complete extraction of cracks.The experimental results show that the proposed method can eliminate the interference of the groove and keep the integrity of the crack after splicing.
展开▼
