一种低能见度下红外机场区域识别新方法

针对低能见度下红外图像背景复杂、目标对比度低的特点，提出了一种新的基于感兴趣区提取和区域生长的红外机场区域识别方法。首先，采用自适应Wiener滤波对图像进行预处理，以削弱图像的背景杂波并增强机场区域的信噪比；然后，利用机场区域和背景的灰度分布特性的差异，在图像预处理的基础上采用自适应双阈值分割和自适应局部极差阈值分割相融合的方法实现机场区域的初步分割；其次，利用机场区域的形状约束和长宽比特征，采用形态学处理和连通域标记实现感兴趣区域的提取；最后利用有限约束的区域生长实现机场区域的识别。该方法结合了感兴趣区域提取和区域生长的优势，能够以较少的计算代价实现机场区域较完整的识别。实验表明，该方法能够有效检测识别机场区域。 

Novel airport region recognition method in low visibility infrared image

Background clutters are usually complicated while contrasts of targets are usually low in low visibility infrared image. On account of this, a new airport recognition method is presented in this paper based on the region-of-interest extracting algorithm and the region growing algorithm. The method is made of four parts. Firstly, image pre-processing is done with adaptive wiener filtering to suppress image background clutters and to improve signal-to-noise of targets. Secondly, image fusion threshold is done applied two threshold methods based on the gray distribution difference between airport and image background clutters, the one is adaptive two-threshold segmentation and another is the adaptive local range segmentation. Thirdly, using airport region shape character and length-width-ratio of airport, the region-of-interest of airport is found via morphology processing and connected-domain-label algorithm. Finally, airport region is recognized applied region growing with prior image constraint. This method takes advantages of the region-of-interest extraction method and region growing method. And it can recognize airport in integrity with lower computational complexity. Some experiments show that the proposed method has good performance of detecting and recognizing airport region in low visibility infrared images.