Abstract: Regarding the characteristic of X-ray detection images of carbon material, flaw feature extraction and selection techniques are studied. Defect style and imaging character of carbon product that easy create in the course of produce are analyzed, based on that, nineteen features are extracted from flaw stylebook. Mathematics model of feature combination classification is regarded as fitness function, optimal selection of original flaw feature is realized with feature selection strategy based on genetic algorithm. Pattern classification of flaw is carried out with BP neural network and the feature selected. Experiment results show that, the method of feature selection is relatively effective, and it could be used for the recognition and classification of flaw.