Image Processing Based Fabric Detection

Abstract

This project titled “Image Processing Based Fabric Detection”. The purpose of this project is to detect fabric. Lack of knowledge about fabric can cause people to have lower quality or wrong type of fabric. Image processing is very complex but we are making progress in this sector. Many difficult problems are being solved by image processing. The utility of image processing in today’s world is vastly accepted and used. Object or type detection is a very important part in image processing. In this research we have developed a method that can detect two of the most common types of fabrics – Cotton and Polyester using the available methods of classical image processing. A simple image takes from the smart phone can tell them the right type of fabric with high accuracy. Our method is trained with many images. The method uses decision tree based feature selection method to find important features of an image out of large feature space and SVM to classify image with the knowledge from the training data. We have obtained ~90% accuracy in successfully separating two above mentioned types of fabrics from cell phone images.