The feature descriptor of Pyramid Histogram of Words (PHOW) was studied,and an aerial reconnaissance image classification method was proposed based on PHOW feature descriptor and Support Vector Machine (SVM) classifier.The experimental results on five kinds of aerial reconnaissance images show that,compared with the image classification method based on Gist feature descriptor,the new method is able to realize image classification rapidly and precisely.The average classification precision is about 84.6%,and the time used for feature extraction and classification is about 37 ms.