摘要
由于现代军事斗争准备的需要，军队要求每一名军人必须了解各种武器装备的性能，这是在作战中取胜、减少人员伤亡的必由之路。然而，世界各国武器多种多样，把所有种类的武器装备都拿来演示是不可能的。这样，使用标有大量信息的军事图片来研究各种武器装备的性能就成了当务之急。因此，有效的图像信息检索技术在军事上就有了极其重要的作用。然而，目前对图像信息的管理技术可以说是基于像素的，而不是基于图像内容理解的。随着军事图像数据量的剧增，仅靠常用的文本标识、关键字等进行索引，局限性太大，无法进行直接基于视觉特征的检索。因此，必须提供有效的图像分析和检索机制，使图像管理和检索高效、易行，这使得基于内容的图像处理技术的研究成为必然。这种技术的优点是在不要求理解图像的前提下充分利用其内容的一些可计算特性，诸如颜色、纹理、形状等，结合其它一些现有的成熟技术，来对图像信息进行存储、管理和检索。
由于现在还有很多技术难题没有得到解决，因而现在投入使用的基于内容的图像检索系统还很少，特别是没有适合于军事武器装备图像检索方面的系统。但在军事上对这种系统的应用需求还很迫切，本文就是在这种前提下，针对军队数字图书馆应用的需求，对基于内容的图像检索技术进行了一些有益的探讨。首先在国际通用的元数据标准Dublin Core的基础上制定了适合军队要求的图像元数据集，并详细分析了颜色、纹理、形状等视觉特征的提取和表示方法；接着探讨了图像视觉特征相似度量的问题，将模糊技术引入到颜色直方图的距离度量中并进行了相关性能的分析比较，指出了几何空间距离度量函数的不足之处，改进了系统中采用的距离函数；然后针对图像视觉特征向量的多维特性，分析了现有的各种降维技术和多维索引技术。选取了适合系统的降维技术和建立多维索引的方法。最后综合以上技术，设计实现了一个综合的基于内容的图像检索系统，集成了元数据检索、全文检索、基于视觉特征检索等多种检索方法，为用户提供了友好的使用界面，本系统为军队使用军事图片研究各种武器装备的性能提供了一种有效的检索机制，大大提高了检索效率。
Modern military conflicts require that every soldier get acquainted with the performances of the varied weapons and equipments, which guarantees the victory and light casualties. However, it's impossible to demonstrate all the weapons and equipments in the world. Recently, weapon study through information-heavily-loaded military pictures and photos becomes an effective alternative. Therefore, image information retrieval (IIR) is increasingly applied in military. Given that present image information management is still based on pixels instead of contents, with the sharp increase of the military image data, the traditional approach of IIR solely based on texts and key words exposes its limitations and cannot be applied directly to the vision-based retrieval. To make image management highly effective and easy to operate, we need an efficient image analysis and retrieval mechanism, i.e. content-based image processing technology. The advantage of this technology is that combined with certain existing techniques,
we can make full use of the calculable features of an image, such as color, texture and shape, to store, manage and retrieve the image without full understanding of it.
Owing to some unsolved technological problems, few content-based IIR systems are put into use, especially in the field of military image retrieval. This paper is an explosive discussion on content-based IIR in accordance with military digital library (MDL). Based on the international general metadata standard Dublin Core, we first established the military image metadata, and then analyzed the extraction and description of the visual features such as color, texture and shape. Next we discussed the similarity measurements of visual features, imported fuzzy logic into the distance feature, compared correlative performances, pointed out the disadvantages of space-based geometry functions, and brought forward a distance function. For multi-dimension vector's high dimension nature, we probed into the existing dimension-lowering technology and multi-dimension index technology, and selected the most suitable method to establish multi-dimension index in our system. Combing the available IIR technologies, we designed
a comprehensive contend-based IIR system with friendly interface for users, which implemented metadata retrieval, text retrieval and vision-based retrieval. This system offers an effective and efficient retrieval mechanism for the Army to study all the available weapons and equipments with military images.
引文
[1] Sclaroff Setal, ImageRover. a content-based image browser for the World-Wide Web. Proceedings of IEEE Workshop on Content-Based Access of Image and Video Libraries. San Juan, Puerto Rico. June 1997. 2-9
[2] Eakins J P, Graham M E and Boardman J M. Evaluation of a trademark retrieval system in 19th BCS IRSG Research Colloquium on Information Retrieval. Robert Gordon University, Aberdeen, electronic. Workshops in Computing, Springer-Verlag, Berlin. March 1997. 36-39
[3] Flickner, Metal. Query by image and video content: the QB1C system. IEEE Computer 28(9), 1995. 23-32
[4] 李国辉，胡晓峰．基于内容的检索．计算机世界报技术专题，1998，3．9-10
[5] 方艳梅，杨灿，肖白美．支持基于内容检索的图像数据模型的设计．计算机应用，2001，7．16-18
[6] 高文，刘峰，黄铁军等著．数字图书馆——原理与技术实现．第一版．北京：清华大学出版社，2000．10．218-222
[7] 唐立军，段丽娟等，基于内容的图像检索系统．计算机应用研究，2001，7．23-25
[8] Ramesh Jain. Infoscopes: information Systems for the Next Century. Proceedings of Multimedia Information System and Hypermedia. Tokyo, Japan, March 1995. 65-66
[9] Pentland. R.W. Picar, S. Sclaroff. Photobook: Tools for content-based manipulation of image databases. SPIE Symposium on Electronic Imaging Science and Technology: Storage and Retrieval for Image Video Databases. V2185, San Jose. CA. 1994. 45-47
[10] 鲁东明，潘云鹤，张畅．基于色彩和线描特征相似度计算的敦煌壁画检索．智能计算机接口应用进展——第三届中国计算机智能接口与智能应用学术会议论文集．1997，9．104-106
[11] Weibel, Stuart. Dublin Core and the Metadata Landscape: Conventions for Semantics, Syntax, and Structure in the Internet Commons. 1999. 19-20
[12] 曹丽华．基于内容检索中的视频处理技术研究．计算机工程与应用，1998，6．43-44
[13] G. Salton and M.J. McGill. Introduction to Modern Information Retrieval. McGraw-Hill Book Company. July 1983. 56-60
[14] John R. Smith and Shih-fu Chang. Quad-Tree Segmentation For Texture-based Imag Query. in ACM Multimedia' 94.1994. 71-72
[15] Rohini K. Srihari. Automatic Indexing and Content-Based Retrieval of Captionaed Images. IEEE Computer 1995. 49-56
[16] 金韬，任秀丽．图像检索中颜色特征的提取与匹配．计算机辅助设计与图形学学报，2000，6．29-30
[17] R. Picard. T. Kabir, and F. Liu. Real-time Recognition with the Entire Brodatz Texture Database. IEEE Conf. On Comp. Vis. And Pattern Recognition. New York, 1993.638—639.
[18] R. M. Haralick. Texture features for image classification. IEEE Trans. Syst, Man and Cybern, SMC-3[6]: 610—621. 1973
[19] C. Dorai and A.K. Jain. Shape spectrum based view grouping and matching of 3d free-form
objects. IEEE PAMI, 19(10): 1139—1146. 1997
[20] V. Gudivada and V. Raghavan. Design and evaluation of algorithms for image retrieval by spatial similarity. ACM Transactions on Information Systems, 13(2): 115—144, April 1995.
[21] Yong Rui. Alfred C. She. and Thomas S. Huang. Modied fourier descriptors for shape representation: a practical approach. In Proc of First International Workshop on Image Databases and Multi Media Search 1996.49-50
[22] Tversky. Amos (1977). Features of Similarity. Psychological Review, 84(4). 327-352
[23] 沈清，汤霖．模式识别导论．长沙：国防科大出版社．1990，8．110-114