In this paper, we explore the problem of content understanding of public events images, and our objective is preventing malicious utilization of public events image in the Internet to impose negative effect on the public opinion. Towards this goal, we propose a public events image content understanding approach, which is based on the object combination feature. We first identify the typical objects of the public events image, then employ SVM to classify the public events image according to the image content and finally succeed to recognize the public events image. Experiments validate the effectiveness of our approach.