基于均衡化与同态滤波的光学图像去云模型
2019-05-06费子豪刘勇李虎吕鑫
费子豪 刘勇 李虎 吕鑫



摘 要:近年来对太阳耀斑的研究受到大多数研究人员的青睐,但获得的影像资料容易受到云层影响。在图像处理中,云雾的存在不仅使太阳耀斑信息被削弱,影响准确性,而且对太阳耀斑的分析、处理、识别和预测造成严重影响。因此,太阳耀斑图像中云雾的去除显得十分必要,对后期的研判、分析及预测都有着重要的意义。首先对原始图像进行像素值提取,并进行高斯降噪、椒盐降噪处理和小波分析,分解高低频信号。在此基础上,建立了基于高频分量直方图均衡化,低频分量同态滤波处理的小波变换模型来进行去云处理。同时,将新模型与未经分解的图像进行同态滤波、小波去云进行对比去云效果,并对其进行分析,组合去云处理的去云效果较明显。针对该算法,通过改变反射参数、边缘锐度等4个指标进行鲁棒性检验。
关键词:OpeCV处理 同态滤波 小波重构 鲁棒性分析
Abstract: In recent years, research on solar flares has been favored by most researchers, but the images obtained are susceptible to cloud imagery. In image processing, the presence of clouds not only weakens the solar flare information, but also has image accuracy, and it has a serious impact on the analysis, processing, recognition and prediction of solar flares. Therefore, the removal of clouds in the solar flare image is very necessary, and it has important significance for the later research, analysis and prediction. Firstly, the original image is extracted with pixel values, and Gaussian noise reduction, salt and pepper noise reduction processing and wavelet analysis are performed to decompose the high and low frequency signals. On this basis, a wavelet transform model based on high-frequency component histogram equalization and low-frequency component homomorphic filtering is established to perform de-cloud processing. At the same time, the new model and the undecomposed image are homomorphic filtered, and the wavelet is compared with the cloud to compare the cloud effect, and the cloud effect is combined. For the algorithm, the robustness test is carried out by changing the four parameters of reflection parameters and edge sharpness.
太阳耀斑(Solar flare)是指发生在太阳表面局部区域中突然和大规模的能量释放过程[1]。它是空间环境的主要扰动源, 对地球空间环境造成很大影响。平均而言,一个正常发展的黑子群(Sunspot groups)几乎几小时就会产生一个耀斑。
近年来对太阳耀斑的研究也受到大多数研究人员的青睐,但是在观测时受到云的遮挡,拍到的耀斑图片需要经过一定的处理才能被用作研究。
去云处理(Go cloud processing),一般是通过对同一地区不同时间的多幅光学图像进行融合[2-3],消除云覆盖,而多幅图像的取得并不容易。……
