Target classification is a key link in the field of computer vision and pattern recognition. SVM(support vector machine) is a new machine learning method put forward based on statistical learning theory. In this paper, an offline image target classification algorithm based on gradient histogram feature of support vector machine is proposed. First, the training set is preprocessed, and then the image is extracted by histogram feature extraction. Finally, the classifier can be detected by training. The test images are tested by using the classifier. The test results show that the target classification test has good effect.