Classification of Building Images using Fractal Features

Abstract

Cracks in concrete buildings may show the total extent of damage or problems of greater magnitude. Causes of cracks depend on the nature of the crack and the type of structure. Crack classification is an approach to using machine learning algorithms to find a particular type of crack. The image is preprocessed by image smoothening and removes noise using a Gaussian filter, whereas the Sobel edge detection method is used to detect the edges. By using k-means clustering, the image segmentation is carried out to identify the Region of Interest. Fractal dimension is an efficient measure for complex objects. Fractal features like fractal dimension, average, and lacunarity are calculated using a differential box-counting algorithm. The classification of the crack classifies the crack based on the characteristics derived from the crack area.