Görüntü işleme ile çatlak incelemesi

Cracks can damage the concrete's structure. Large structural problems can occur in concrete structures when the cracks expand. Crack detection can avoid the main problems. This project puts forward a new technique of crack detection in the processing of images. The proposed crack detection method generally consists of four steps. Converting input image to a gray-scale image and then using Gauss Filter and Median Filter for removing noise from a gray-scale image is first step. Secondly, the gradient of the first step output image is determined and the maximum non-suppression is applied to the gradient image and then mapped for noise removal. Thirdly; the binary image is obtained from the second step output image and the binary image is applied with a morphological close operation. In the fourth, the crack detection threshold is calculated to decide whether the image input has a crack or not. For two hundred concrete sample images with one hundred non-crack and one hundred crack images, the accuracy of the proposed method is equal to ninety-six percent. You can accept this method that the images are not cracks or cracks. This method can be used to find sample images have cracks or not.