A Method for Global-scale Archiving of the Imaging Data Based on QTM Pixel

Wenbin SUN 1 Xuesheng ZHAO 1,2 Jun CHEN 2

1 Department of Surveying and Land Science , China University of Mining and Technology ( Beijing ), Beijing , China , 100083

2 National Geometrics Center of China , No.1 Baishengcun, Zizhuyuan, Beijing , China , 10004

To efficiently utilize natural resource and monitor the environmental change, the seamless management and operation of the image data on the global-scale is urgently needed. The longitude/latitude grid is often used with its structure of the continuity and hierarchy on the globa1scale in order to avoid the problems of data incontinuity or overlapping caused by planar map projection [Mulcahy, 2001]. However, the efficiency is very low in the multi-resolution data management and operation as the area and shape of the longitude/latitude grids varies enormously with changing of latitude, especially in the areas of higher latitude. The QTM (Quaternary Triangular Mesh, Dutton 1989) as a continuous, hierarchal quadtree data structure with the uniform grids on sphere is one of the most efficient methods for managing the global spatial data in many applications. But as so far, it has been seldom approached in an expression of the global imaging data seamlessly. In this paper, a new method for global-scale archiving of the imaging data based on QTM, will be presented.

Firstly, the mapping relation between image pixel and QTM grid is constructed based on the QTM subdivision and its coding scheme (Quaternary coding, approached by Bartholdi [2001]). The main principle includes two parts: the first part is the conversion process between QTM code and longitude/latitude coordinate, i.e., according to the properties of the QTM coding scheme, the rules of triangular direction can be defined by the number of “1” in the quaternary code. Therefore, there are four cases in the process of conversion between the code of QTM and longitude/latitude according to triangle direction and parity of “3” number in its code. The conversion between longitude/latitude of sphere and QTM code can be finished by using the principle of Equal Triangles Projection (approached by Goodchild[1992] ). The second part is the construction process of mapping relations based on the sub-pixel-method, i.e. determining the level of QTM grids by the size of original imaging pixel. The detail steps are as follows: at first, the sum of QTM grids, which is corresponding to original imaging pixels, is calculated by the transformation from the coordinate of midpoint of sub-pixel to QTM code. Then, by eliminating the duplicate code and edge grid, the remainder is the corresponding mapping grids in QTM.

Secondly, a corresponding algorithm of QTM pixel calculation is developed. In this method, the area is considered as a main affecting factor. The principle is as follows: at first, three vertexes of the QTM grid are projected to the projection plane of original imaging data and an approximate planar triangle is formed. The grey level of QTM grid can be calculated based on the weight of the overlapping area between this triangle and original imaging pixel on the planar. After that, the global imaging data are segmented and managed by lower resolution grid based on the QTM hierarchal subdivision. Next, The size of each segment is determined by the level of QTM. The data in one segment is stored as BLOB format and indexed by Quaternary code. Therefore, the quick query of remote sensing data can be realized by calculating the QTM code in the corresponding spatial range of query.

In the end, the experiment is given by using the 1km resolution NOAA data in china. In order to compare the difference of grey level between original imaging pixels and QTM pixel, 100000 points are randomly selected from the range of rectangle encircled by 81 º 53 ¹ 3.71 ″ E,88 º 22 ¹ 11.32 ″ E longitude and,45 º 10 ¹ 44.66 ″ N,42 º 12 ¹ 44.00 ″ N latitude. The result indicates that:

• The conversion from imaging pixels to QTM pixels can be quickly finished by using the algorithm of mapping relations.
• The global imaging data archiving and managed based on QTM is continuous, hierarchal and uniform without the problems of geometry distortion caused by longitude/latitude grid and the data incontinuity caused by planar map projection.
• The accuracy loss of conversion from the imaging pixel to QTM pixel is that: the loss less than 2 grey level is 94.5% of all pixels, the loss from 2 to 4 is 1.9%, the loss from 4 to 10 is 2%, the other is 1.6%. The results are good and receivable.

Keyword: Seamless Archiving QTM Pixel Grey Level Remote Sensing Data

References

Mulcahy K. A.. Map Projections and Spatial Referencing For Global Data Sets. http://www.discreteglobalgrids. -org. 2004

Dutton G. 1989. Modeling locational uncertainty via hierarchical tessellation. In: Accuracy of Spatial Databases. London: Tayor and Francis. 125-140

Bartholdi J.J., and Goldsman P. 2001, Continuous indexing of hierarchical subdivisions of the globe. International Journal of Geographical Information Science. VoL.15, No. 6, 489-522