4. Conclusions
This paper proposes a new crack defects detection algorithm that combines the fast discrete curvelet transform method and the texture feature measurement method. The gray level co-occurrence matrix is used to analyze the second-order statistics feature in the reconstructed image; the right functions are selected to obtain the right measures for the reconstructed image; the best choice of the ratio of the remaining coefficients is determined through the texture feature measurement, and the threshold of the reconstructed coefficients can be calculated automatically. The texture in the original image can be removed by curvelet reconstruction with the reset threshold and in the reconstructed image the crack defects can be extracted correctly without the influence of its grinding texture. Compared with traditional method (the Otsu algorithm and the morphological filtering algorithm), the algorithm in this paper is more effective and feasible.
It should be mentioned that, although the effectiveness of the proposed technique has been demonstrated, detection scheme requires the light and the field of view strictly, and diffuse reflection and bright field are needed to make the cracks evident. Researchers have proposed many algorithms for illumination dependence [27] and [28] and achieved good results, we will try to use these methods to achieve invariant lighting in the next phase, and then the robustness of the system can be further increased. Accuracy rate of up to 93.9% can be achieved with the proposed approach and the length of the cracks greater than 0.8 mm can be extracted accurately.
Acknowledgments
This research is supported by National Key Technology Research and Development Program of the Ministry of Science and Technology of China (No. 2012BAF06B02), and National Key Technology Research and Development Program of the Science and Technology Department of Sichuan Province of China (No. 2011CGZ0049).
4. Conclusions
This paper proposes a new crack defects detection algorithm that combines the fast discrete curvelet transform method and the texture feature measurement method. The gray level co-occurrence matrix is used to analyze the second-order statistics feature in the reconstructed image; the right functions are selected to obtain the right measures for the reconstructed image; the best choice of the ratio of the remaining coefficients is determined through the texture feature measurement, and the threshold of the reconstructed coefficients can be calculated automatically. The texture in the original image can be removed by curvelet reconstruction with the reset threshold and in the reconstructed image the crack defects can be extracted correctly without the influence of its grinding texture. Compared with traditional method (the Otsu algorithm and the morphological filtering algorithm), the algorithm in this paper is more effective and feasible.
It should be mentioned that, although the effectiveness of the proposed technique has been demonstrated, detection scheme requires the light and the field of view strictly, and diffuse reflection and bright field are needed to make the cracks evident. Researchers have proposed many algorithms for illumination dependence [27] and [28] and achieved good results, we will try to use these methods to achieve invariant lighting in the next phase, and then the robustness of the system can be further increased. Accuracy rate of up to 93.9% can be achieved with the proposed approach and the length of the cracks greater than 0.8 mm can be extracted accurately.
Acknowledgments
This research is supported by National Key Technology Research and Development Program of the Ministry of Science and Technology of China (No. 2012BAF06B02), and National Key Technology Research and Development Program of the Science and Technology Department of Sichuan Province of China (No. 2011CGZ0049).
การแปล กรุณารอสักครู่..
