设为首页
加入收藏
基于模糊纹理光谱的全天空红外图像云分类
Cloud Classification of the Whole Sky Infrared Image Based on the Fuzzy Uncertainty Texture Spectrum
-
摘要: 为了对全天空红外测云系统获得的红外图像进行云类自动识别, 提出了基于模糊纹理光谱结合云物理属性的全天空云类识别方法。首先根据不同滤波窗口的模糊纹理光谱图像特征, 确定了滤波窗口大小, 然后通过分析不同天空类型下的FUTS谱 (fuzzy uncertainty texture spectrum) 以及同一种天空类型下的FUTS谱, 考察了FUTS进行云类识别的适用性, 最后利用最小距离分类法和云基本物理属性对全天空红外图像进行了分类测试。在200个测试样本中, 层状云、积云、高积云、卷云和晴空的识别率分别为100%, 100%, 90%, 100%, 100%, 平均识别率达到98%。基于模糊纹理光谱的云分类算法对单一云空具有很好的分类效果, 可进一步应用于全天空红外图像的云分类识别。Abstract: Clouds play an important role in the earth radiation budget and climate change. Their shape, size, distribution and movement indicate the condition of the atmosphere.Nowadays, cloud amount and cloud height are collected by means of both satellites and ground-based instruments. Satellite cloud images provide global coverage, and these data are used widely in weather forecast. Ground-based cloud images are very local ones which contain more details of clouds.Cloud classification using satellite images has been done for many years, while the study of ground-based cloud classification is still underway. A method using fuzzy uncer tainty texture spectrum and essential information in cloud images is proposed to classify five sky conditions (stratus, cumulus, altocum ulus, cirrus and clear sky) autom atically based on cloud images obtained from the whole sky infrared cloud measuring system (WSICMS).The WSICMS is a ground-based passive sensor that uses an uncooled microbolometer detector array to measure downwelling atmospheric radiance in the 8—14μm wavelength band of the electromagnetic spectrum. It provides a way to identify clouds, obtain clouds distributions and calculate clouds amounts continuously with no difference in sensitivity during day and night. The primary WSICMS components are optical detector, environmental parameter sensors, controller and power component. The optical detector is an uncooled microbometer array containing 320×240 pixels. It obtains nine images at zenith and at each eight orientations under the control of the scan servo system. A whole sky image is accomplished after spelling nine images, water vapor correction and zenith angles correction.The WSICMS locating at Nanjing, China has been working since August 2006. The 200 cloud images according to human observations are selected randomly from these sample sets. Before cloud classification, an appropriate FUTS filter window (7×7) is chosen. Analyses of FUTS of five different sky conditions and same sky condition (cumulus) show that FUTS can serve as a good discriminating tool in cloud classification. Based on above analysis, a supervised classification with minimum distance rule is used to classify sky conditions. The classification accuracy rates of stratus, cumulus, altocumulus, cirrus and clear sky compared with human observations increase sharply after adding essential information in cloud images. Importance of the cloud characteristic is shown in cloud classification. The final classification result are 100%, 100%, 90%, 100% and 100% respectively, the average accuracy rate is 98%. Altostratus, cumulostratus and complex sky conditions are not discussed here. Future work on this project will focus on this. In addition, more particular sample sets should be built up to improve the accuracy of both training and test data.
-
图 1 不同滤波窗口下的模糊纹理图像
Fig. 1 The fuzzy cloud images corresponding to different filter windows
图 2 不同滤波窗口下模糊纹理光谱的变化
Fig. 2 Differences of the FUTS in fuzzy cloud images corresponding to different filter windows
表 1 模糊纹理光谱分类混淆矩阵
Table 1 The confusion matrix for the test sky condition
表 2 模糊纹理光谱加灰度统计量分类混淆矩阵
Table 2 The confusion matrix for the test sky condition with the essential information in the cloud image
-
[1] Peura M, Visa A, Kostamo P. A New Approach to Land-Based Cloud Classification//Proceedings of ICPR, IEEE, 1996:143-147. [2] Lohmann G. Co-occurrence-based Analysis and Synthesis of Textures//Proc 12th IAPR Internat Conf Pattern Recognition, Vol 1, Jerusalem, 1994. [3] Haralick R M, Dinstein I, Shanmugam K. Textural features for image classification. IEEE Trans Syst Man Cybern, 1973, 3: 610-621. http://www.academia.edu/963894/Textural_features_for_image_classification [4] Welch R M, Kuo K S, Sengupta S K, et al. Cloud field classification based upon high spatial resolution textural feature (Ⅰ) : Graylevel cooccurrence matrix approach. J Geophys Res, 1988, 93 (10) : 12663-12681. https://www.researchgate.net/publication/23838114_Cloud_field_classification_based_upon_high_spatial_resolution_textural_features_I_-_Gray_level_co-occurrence_matrix_approach [5] Kuo K S, Welch R M, Sengupta S K. Structural and textural characteristics of cirrus clouds observed using high spatial resolution landsat imagery. J Appl Meteorol, 1988, 27(11) : 1242-1260. doi: 10.1175/1520-0450(1988)027<1242:SATCOC>2.0.CO;2 [6] 傅德胜, 王新芝.云图纹理特征的抽取与云的自动分类.南京气象学院学报, 1995, 18(4):530-535. http://www.cnki.com.cn/Article/CJFDTOTAL-NJQX504.010.htm [7] 寿亦萱, 张颖超, 赵忠明, 等.暴雨过程的卫星云图纹理特征研究.南京气象学院学报, 2005, 2(3):337-343. http://www.cnki.com.cn/Article/CJFDTOTAL-NJQX200503007.htm [8] 夏德深, 金盛, 王健.基于分数维与灰度梯度共生矩阵的气象云图识别.南京理工大学学报, 1999, 2(3):278-281. http://www.cnki.com.cn/Article/CJFDTOTAL-NJLG903.022.htm [9] Lamei N, Hutchison K D, Crawford M M, et al. Cloud-type discrimination via multispeetral textural analysis. Opt En, 1994, 33 (4) : 1303-1313. doi: 10.1117/12.166920 [10] Du L J. Texture Segmentation of SAR Images Using Localized Spatial Filtering//Proc Int Geosci Remote Sensing Symp, Washington D C, May, 1990: 1983-1986. [11] 白慧卿, 方宗义, 吴蓉章.基于人工神经网络的GMS云图四类云系的识别.应用气象学报, 1998, 9(4):402-409. http://qk.cams.cma.gov.cn/jams/ch/reader/view_abstract.aspx?file_no=19980460&flag=1 [12] 周著华, 白洁, 刘健文, 等.MODIS多光谱云相态识别技术的应用研究.应用气象学报, 2005, 16(5):678-684. http://qk.cams.cma.gov.cn/jams/ch/reader/view_abstract.aspx?file_no=20050587&flag=1 [13] Buch K A Jr, Sun C H. Cloud Classification Using Whole-sky Imager Data//Proceedings of the ninth Symposium on Meteorological Observations and Instrumentation, Charlotte, North Carolina, 1995: 353-358. [14] Haralick R M. Statistical and Structural Approaches to Texture // Proc IEEE 67, 1979: 786-804. [15] Lu C S, Chung P C, Chen C F. Unsupervised texture segmentation via wavelet transform. Pattern Recognition, 1997, 30 (5) : 729-742. doi: 10.1016/S0031-3203(96)00116-1 [16] He D C, Wang L. Texture unit, texture spectrum and texture analysis. IEEE Trans Geoscience and Remote Sensing, 1990, 28 (4):509-512 doi: 10.1109/TGRS.1990.572934 [17] Lee Y G, Lee J H, Hsueh Y C. Fuzzy uncertainty texture spectrum for texture analysis. Electron Lett, 1995, 31 (12) : 959-960. doi: 10.1049/el:19950665 -
下载:
图(5) / 表(2)
计量
- 摘要浏览量: 3985
- HTML全文浏览量: 692
- PDF下载量: 1823
- 被引次数: 0
