This article presents a two-stage approach, combining novel and traditional algorithms, to image segmentation and defect detection. The first stage is a new method for segmenting fabric images is based on Hamiltonian quaternions and the associative algebra and the active contour model with anisotropic gradient. To solve the problem of loss of important information about color, saturation, and other important information associated color, we use the quaternion framework to represent a color image to consider all three channels simultaneously when segmenting the RGB image. In the second stage, our crack and damage detection method are based on a convolutional autoencoder (U-Net) and deep feature fusion network (DFFN-Net). This solution allows localizing defects with higher accuracy compared to traditional methods of machine learning and modern methods of deep learning. All experiments were carried out using a public database with examples of damage to the TILDA fabric dataset.
|