摘要
红外成像自动目标识别是精确制导武器的重要研究内容。针对地面复杂背景条件下,目标难以检测识别的问题,在分析地面目标红外图像场景的基础上,首先利用形态学技术对图像进行去噪和抑制背景,采用Otsu法对图像进行阈值分割并用形态学技术消除虚警目标点,用扩展像素标记法对图像进行区域标记,然后提取目标形状、矩和统计分布三类特征作为识别特征,并采用模糊综合评判方法识别目标。对实地拍摄的100幅坦克目标红外图像进行了识别实验,结果表明,该方法在坦克的典型作战环境下具有较强的抗噪性能和抑制背景结构干扰的能力,能有效识别出目标,并且易于用并行处理和硬件实现。
Automatic Target Recognition （ATR） through infrared imaging is an important part in the research of the precision guided weapons. Since it is difficult to recognize targets with complex background, we analyzed the IR images of targets on the ground, used morphology technique to reduce the noise and restrain the background. Then we used Otsu algorithm for threshold segmentation and used morphology technique to remove the targets of false alarm. Extended pixel label method was used to label regions, then target shape, moment and statistieal distribution were extracted as recognition characteristics, and fuzzy compositive evaluation method was used for target recognition. 100 IR images of tank were used in the recognition experiment, the result showed that the method used here has the ability of anti-noise and inhibiting background interference, which can recognize the target effectively and is easy for parallel processing and hardware implementation.
出处
《电光与控制》
北大核心
2009年第3期71-74,共4页
Electronics Optics & Control
关键词
红外目标识别
自动目标识别
形态学滤波
图像分割
IR target recognition
automatie target recognition
morphological filter
image segmentation
