地基云图识别的光照补偿算法

为研究地基云图云状识别中的光照补偿问题，针对可见光地基云图，提出了一种基于BEMD与Closed-Form景物提取算法相结合的地基云图光照补偿算法对云图进行预处理，经过去光照处理后的云图，再进行Hilbert谱特征提取，最后进行云类的识别。与基于L1的全变分模型的光照补偿算法相比，可有效地去除云图的光照影响，又不至于去掉云图自身的信息，且与背景区别明显。试验结果表明，使用该方法与未进行处理的云图相比识别率提高4.17%~18.88%。

De-illumination method about recognition of ground-based cloud

To study the de illumination of automatic cloud type classification, a cloud de illumination algorithm was presented based on BEMD combined with a closed form solution to natural image matting for visible ground nephogram. The feature of Hilbert spectrum was extracted from de illumination cloud image. Then the processed cloud images were used to do cloud classification. This method has good de illumination performance without losing its value information and with clear differences from the background compared with the algorithm based on L1 of total variation mode. In addition, a test run of random images was presented, still outperforming previous algorithms by yielding a success rate, a rise of 4.17%~18.88% percentage points.