Abstract: A fast target recognition algorithm for aerial images with large obliquity is presented，which especially fits to the case that the targets are plane targets on the earth.The method comes over the deficiencies，such as the complexity of the models，the load of computational complexity，and the lack of the real time property，to a large extent.At the same time，it improves the precision of target recognition.The algorithm firstly detects the interest regions that may contain targets and rectifies the regions；then recognizes the targets form the rectified regions；finally outputs the information of targets recognized form the regions to complete the recognition of the targets.The experimental results show that image rectification is very promising and potentially useful for target recognition.

Key words: target recognition, image rectification, aerial image, large obliquity

摘要： 提出了一种基于图像校正的快速目标识别算法，特别适合于航拍照片中地面上面目标的识别。算法可以使目标模型单一化，在很大程度上克服了目标识别模型复杂、数据运算量大、计算实时性差等缺点，提高了目标识别的实时性和精确性。首先对图像进行感兴趣区域的检测；对检测出的区域进行图像的正视校正；然后对校正后的区域进行特征提取；最后进行目标识别，并输出目标信息，完成识别过程。实验表明，该算法用于大倾角航片目标识别是有发展潜力和前途的。

关键词: 目标识别, 图像校正, 航片, 大倾角