红外图像上的单散雷探测

Single Landmine Detection in Infrared Image

针对红外图像中单散雷的探测问题,提出了基于图像的随机模型及信号检测理论的地雷探测方法。从自然背景红外图像的随机描述出发,将背景红外图像建模为自回归(AR)模型,并用现代信号处理的方法将其转化成高斯白噪声,再利用信号检测理论中的Rao检验进行地雷信号的检测,解决了复杂背景下红外图像中单散雷探测及定位的难题。较之现有的匹配滤波检测方法,该方法有着更好的检测性能,试验结果也验证了该方法的优越性。

For the question of single landmine detection in infrared image, a new way was given, which based on the stochastic model of image and the signal detection theory. According to the stochastic description of the background image, we can modeling the background infrared image with the autoregressive model, and change the background image into white Gauss noise by the modern signal processing method, then use the Rao test to detect the signal of landmine in infrared image, finally we have solved the problem of the single landmine detection and location in the infrared image under the complex background. Compared with the match filter method, this method has the better detection performance, and the results also prove this method superiority.