改进LBP和HSV颜色直方图相结合的地表状态识别

在利用近地表遥感探测方法对地表信息进行反演时，地表状态的不同会导致电磁波传播方式不同，进而决定了所采取的遥感探测方式、探测波长及探测手段的不同。因此，地表状态的识别是进行近地表遥感反演的前提和基础，其识别的准确率也决定了反演结果的准确度。本文提供了一种改进LBP（local binary pattern）和HSV颜色直方图相结合的地表状态识别算法，通过改进阈值的LBP算法和HSV颜色直方图提取特征向量并建立判别条件，最后利用K邻近算法对测试样本与训练样本进行特征匹配，得到识别结果。将野外采集的411张图片多粗随机分组为训练样本和测试样本，结果表明206张测试样本的分类正确率都达98.7%以上，证明了本文算法的有效性。

A combined algorithm of improved LBP and HSV for surface state recognition

Different surface states will lead to different modes of electromagnetic wave propagation, which determines the different detection methods, detection wavelengths and detection methods of remote sensing, when the surface information is retrieved by near-surface remote sensing. Therefore, surface state recognition is the premise of remote sensing retrieval of near-surface, and recognition accuracy determines remote sensing retrieval accuracy. A combined algorithm of LBP(local binary pattern)and HSV color histogram for surface state recognition is proposed. Feature vector extraction is carried out by combining the Improved Threshold LBP algorithm and HSV color histogram, then discriminant condition is established, in the end, the results of recognition can be realized by feature matching between test samples and training samples, based on K-neighborhood algorithm. 411 images collected in the field are divided into training samples and test samples randomly. The correct rate of recognition for 206 test samples is more than 98.7%, It is proved that the algorithm is effective.