The defect inspection is a crucial process for the plasma display panel (PDP) production that significantly influences the quality of final products. In this paper, we propose a defect identification and classification method that extracts and classifies defects using various image analysis techniques. First, we identify defects through binarization of images using Gaussian filter. Then, those defects are classified into seven different types by analyzing geometric characteristics of defects and utilizing a support vector machine (SVM) classifier. The experimental results using separate sets of training and test PDP images obtained from production lines are quite promising. Our method identifies defects effectively enough to be used in the real environment. It also achieves a high correctness in classifying various types of defects.