微弱日盲紫外电晕自动实时检测方法

针对目前商用日盲紫外/可见光双谱段电晕探测仪对故障的判定效率低且容易受噪声干扰问题,本文提出了一种微弱日盲紫外电晕自动实时检测方法。在分析电晕目标和噪声时间域统计特性的基础上,该方法利用电晕目标在时间域连续的特点,首先完成灰度图像二值化、形态学膨胀等预处理,其次将N帧连续图像累加后阈值化,最后进行特征提取,获取电晕位置、面积等特征信息,实现电晕的自动实时检测。在完成设备辐射标定的基础上,可立即回溯得到故障的光子计数参考值。建立了该方法的探测概率、虚警率数学模型。将其在高速数字处理平台TMS320DM642上实现并输入测试视频,结果表明:在典型参数情况下该方法单次检测虚警概率为2.85×10^-5,处理时间小于120 ms,可实现微弱日盲紫外电晕的实时检测。

Automatic real-time detection method of faint solar-blind ultraviolet corona

In order to enhance the corona detection efficiency of commercial SBUV/Visible double spectral corona detect systems and avoid the noise interference, an automatic real-time detection method detecting weak solar blind ultraviolet corona is proposed. Based on the analysis of statistical characteristics of the corona target and noise time-domain, utilizing corona characteristics in continuous time domain, the pretreatments like gray image binarization and morphological dilation are completed firstly, followed by accumulation of the N successive frames, then thresholding. At last the features of corona like location, size and so on are extracted. Upon completion of the calibration of radiation equipment, the corona's photon counting reference value can date back immediately. The mathematical model of detecting probability and false arm rate(FAR) is established. We implement the method and test on a high-speed digital processing platform TMS320DM642 using the video of corona, and the results show that its FAR is 2.85×10^-5, and the processing time is less than 120 ms, indicating that the weak corona can be detected in real-time.