In India sixty percentage of the people are depending on Agriculture. The product of the Agriculture depends on the quality of the seeds. Most of the farmers are unaware of what seeds they are using. They may use defected seeds in their fields. To identify the defective seeds, many techniques have been existed. Among those techniques, LBP and its derivatives are good to classify the defected seeds. However, it fails for low contrasted seed images. To overcome these limitations, in this paper GLBP technique is proposed. In this proposed technique, noise pixel, edge pixel, and intensity value of gray level are used to classify the defected seeds. The experimental results in the form of table and graphs show that the proposed method produced better results than the existing systems.