摘要
根据复杂背景下前视红外机场目标的特点,提出了基于知识的自动目标识别算法,算法的设计以实用性为宗旨,在不影响算法性能的前提下尽量提高其实时性能:通过提取感兴趣区域来缩小处理范围,采用双阈值最大类间方差快速算法快速准确找到分割阈值,采用快速标记算法标记分割图像,利用Freeman链码描述机场的轮廓特征,根据模糊综合评判机制分析机场的目标特征。通过硬件处理平台的仿真验证,该算法对机场目标红外图像有一定的识别能力,且具有较高的实时性。
An AT R algorithm of airport objects in infrared images with complex backgrounds is proposed based on knowledge according to its special characteristics.The design of this ATR algorithm is practicality oriented, reducing the time cost without sacrificing the performance of the algorithm. The amount of processing is reduced by extracting the region of interest, and by using Otsu with double thresholds the thresholds are decided quickly. Quick labeling algorithm is used to label the segmented image, and Freeman chain code algorithm is used to describe the profile of the airport. Features of the airport are analyzed by using Fuzzy integrated judgement method. Via testing on hardware platforms, the algorithm proposed is proved to be capable in detecting airport objects in infrared images, with a high real-time performance.
出处
《红外与激光工程》
EI
CSCD
北大核心
2007年第3期398-402,420,共6页
Infrared and Laser Engineering