A Technique for Glass Defect Detection

Glass is a material which is used in the industry and household. The presence of defects or weaknesses in the glass has serious implications. In a glass substrate, the grey level of defects and background are hardly distinguishable and results in a low contrast image. The primary objective of this paper is to develop a method for detection of defects in a glass surface image such as subtle defect , bubble defect , dirt defect checks or marks defect etc. The paper proposes artificial neural network based methodology to detect the defects in the glass Gray level Concurrence Matrix (GLCM) has been used for feature extraction. The neural network is responsible for making intelligent classification based on observations done for various types of glass defects. Experimental results developed a classification matrix and by which performance goal based on MSE (mean square error) is set. .This technique helps to get better results.