Advanced Materials Research Online: 2014-06-25 ISSN: 1662-8985, Vols. 971-973, pp 1607-1610 doi:10.4028/www.scientific.net/amr.971-973.1607 2014 Trans Tech Publications, Switzerland Discussion on Development of Mineral Identification System Based on IDL Yongfei Che 1,a,Yingjun Zhao 1,Wenhuan Wu 1 1 National Key Laboratory of Remote Sensing Information and Imagery Analyzing Technology Beijing Research Institute of Uranium Geology,Beijing 100029, China a cheyongfei@163.com Keywords: hyperspectral remote sensing; mineral identification module; IDL Abstract. The traditional data processing and analysis method of remote sensing image processing system cannot meet the hyperspectral remote sensing mass data processing and the need of practical application in mineral resources exploration. By studying the systematical analysis and key technology on the hyperspectral mineral information identification module, and analyzing and thinking about the relevant theoretical methods and technical process, carried out the development of hyperspectral mineral information identification module based on IDL and integrated with ENVI software, providing the basic support platform for hyperspectral remote sensing mineral resources exploration. Meanwhile, the existing problems were discussed from the spectral characteristics mechanism analysis of rock and the hyperspectral mineral identification optimization algorithms. Introduction At present, the advantage of hyperspectral remote sensing image-spectrum merging has been applied to the geological field, producing a series of data processing and rock ore information extraction method, and formed a set of conventional process. Among them, the domestic and overseas professional remote sensing image processing system, the most representative of the ERDAS development of ERDAS ERDAS IMAGINE, the United States Exelis Visual Information Solutions company's flagship product ENVI, the Canadian company PCI development of PCI GEOMATICA [1, 2]and the Chinese academy of sciences institute of remote sensing applications independent research and development HIPAS have achieved some success in the processing of hyperspectral data. However, these traditional methods based on the analysis of remote sensing data processing have been difficult to meet in the application of mineral resources exploration, unable to realize the recognition of geological prospecting information effectively. The remote sensing image processing software platform, such as ENVI, PCI, ERDAS IMAGINE has the hyperspectral remote sensing data processing module, but the functions are not comprehensive. Moreover, ENVI software for the lack of special modules of the mineral and rock characteristics information extraction, it just extracted the rocks and minerals information directly by using hyperspectral images. Therefore, it is urgent to develop specially using of rocks and minerals information feature extraction of hyperspectral remote sensing information processing system. Based on the above analysis and the demand, combined with ENVI existing file processing functions, this paper was developed the mineral information extraction of hyperspectral remote sensing processing system, in the form of a plug-in to realize the seamless integration with ENVI remote sensing image processing platform. The main module functions include: spectral All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications, www.ttp.net. (ID: 130.203.136.75, Pennsylvania State University, University Park, USA-05/03/16,14:00:39)
1608 New Technologies for Engineering Research and Design in Industry characteristics analysis and mineral information extraction, which offered a new way and the implementation approach for rock and mineral information extraction. Analysis of Hyperspectral Mineral Identification Module Technical Process Analysis. In the process of systematic analysis, using the IDL language development hyperspectral mineral information identification module and seamlessly integrated with the ENVI platform become the special processing module of geological prospecting in ENVI. Based on the actual needs of hyperspectral remote sensing mineral information extraction, the main technological process of the system was designed (Fig. 1). Fig. 1 The System Function Design Functional Analysis. Based on the demand analysis and the technical process analysis, in the IDL language development environment, combining with ENVI file processing functions independently developed the hyperspectral mineral information identification module. In order to achieve these objectives, the analysis of the specific functional requirements as follows: (1)The spectral characteristics analysis sub-module. Analysising of the mineral spectral absorption width, the spectral absorption area, the spectral absorption depth, the spectral slope, the trough position and number, the peak position and number such as characteristic parameters and combined with chemical analysis parameter,construct spectrum relation model. (2) The minerals information extraction sub-module. Using spectral Angle mapping (SAM), spectral correlation coefficient mapping (SCM), spectral information divergence, spectrum waveform matching and spectral feature fitting such as recognition algorithms identified the mineral information. Data Input and Output Analysis.The hyperspectral mineral information identification module is relying on the ENVI platform running, so the data input and output is to use the existing mature technology of ENVI to deal with. Therefore, in this paper the key to consider data input and output is how to connect ENVI data input/output interface with the data processing interface of module, and make them in good state of compatible. Key Technology The main contents of the module function implementation stage include the implementation of the physical system, the program design and debugging and the algorithm implementation, etc. The implementation of the physical system generally refers to the implementation of the computer
Advanced Materials Research Vols. 971-973 1609 hardware and software environment and network system. The developed module is a Stand-alone operation mode without network supporting in this paper, so it does not consider the implementation of the network system.in addition,the ENVI + IDL is not high to the requirement of computer hardware and software environment, the current mainstream of graphic image processing of hard and soft system can meet the operation of the software. Therefore, the implementation of the physical system is not be more discussion in this paper, and the emphasis is on two aspects of program design and debugging and algorithm implementation. Program Design. Automatically adding method is used to add customize menu. Using ENVI_DEFINE_MENU_BUTTON function adds a new menu item. Use this function should be noted that in the process of defining a new menu item file, must contain a named filename define buttons, the process of the new menu item is defined in the process, the filename is the program file file name. Using this function should be noted that Defining a new menu item program files must contain a named "filename_define_buttons" process, a new menu item on the definition in this process, where "filename" is the program files names. Data input and output part use ENVI's own functions to achieve in the process of module running, so we have to consider the design of interface between the module and the ENVI platform. Also said, how to open files with ENVI transferred to self-developed modules internal for processing, and the processed data how to save or export by ENVI. ENVI provides a series of file management functions and subroutines for secondary development, and using the document management functions and subroutines and their uses in this paper were summarized as follows: ENVI_GET_FILE_IDS(), this function returns an array containing the currently ID of each open file, and the file ID is a long integer greater than 0.If currently it didn't open the file, it returns 1; ENVI_FILE_QUERY() returns the data file information, including the row number and column number, the band number,the space dimension, the file name, the band name, the data type, etc; ENVI_GET_MAP_INFO() gets the data file of geographic information, etc; ENVI_ENTER_DATA,this subroutine will be transformed array into images, stored in memory calls in ENVI. ENVI_ENTER_DATA calls ENVI_SETUP_HEAD in the process of running, automatically create a header file and returns a file ID of the image memory. Band information has been added to the Available Bands List, which can be used as the use of any other ENVI images. Algorithm Implementation. Based on hyperspectral images with the characteristics of high spectral resolution, and a large amount of data, the SAI algorithm was designed with the ENVI image partition technology[3] and IDL matrix technology[4] in this paper. First of all, using the ENVI image partition technology the image can be divided into several sub-blocks, and reduces the amount of data reading into memory each calculation. A sub-block for (DIMS (2)-DIMS (1) +1) x NB two-dimensional matrix, DIMS is a five elements of long integer array, which defines the Spatial subsetting of the processing data, and NB is the number of bands. Secondly, the (DIMS (2) -DIMS (1) +1) x NB calculated for SAI. If using the traditional FOR loop through each pixel location (DIMS (2)-DIMS (1) +1), extracting its spectral curve (total NB matrix elements) of SAI calculations, this algorithm can only calculate a pixel location like SAI each access, and the computational efficiency is very low.idl is a computer language based on matrix. The data operations on all matrix elements rather than through cyclic matrix each element reduce the difficulty of interaction analysis, reducing the program running time. Using the IDL matrix
1610 New Technologies for Engineering Research and Design in Industry computation technology, an access can calculate the entire sub-block for SAI that DIMS(2)-DIMS(1)+1 pixels as SAI. The calculation efficiency is higher DIMS (2) -DIMS (1) +1 times than the FOR loop and with the size of DIMS (2)-DIMS (1) +1 continuing to improve. Issues and Discussion In this paper the development of hyperspectral mineral information identification module based on IDL and integrated with ENVI software provide the basic support platform for hyperspectral remote sensing mineral resources exploration. However, the module is just the prototype of the automatic identification of hyperspectral mineral, based on a combination with ENVI software secondary development and IDL, in the process of system development and integration, and it still exists the following problems that continue to improve and perfect. (1) For the spectral characteristics mechanism analysis of rock ore, especially the spectral characteristics research of mixed minerals both in theory and experiment need to be further strengthened and enhanced. This will directly affect the rock classification of the mineral resource exploration application, recognition algorithm design and application effect. (2) Strengthen the optimization algorithm of different mineral, how to adjust the parallel algorithm and design ideas based on the different application requirements and selecting the corresponding parallel implementation platform (such as cluster, GPU) still need to be further perfect. References [1] Boardman J W, Kruse F A. Automated spectral analysis: a geological example using AVIRIS data, north Grapevine Mountains, Nevada: Proceedings of the Thematic Conference on Geologic Remote Sensing, 1994[C]. Environmental Research Institute of Michigan. [2] Van Der Meer F, Bakker W. CCSM: Cross correlogram spectral matching[j]. International Journal of Remote Sensing, 1997,18(5):1197-1201. [3] Xie H, Hicks N, Randy Keller G, et al. An IDL/ENVI implementation of the FFT-based algorithm for automatic image registration[j].computers & Geosciences, 2003,29(8):1045-1055. [4] Marschallinger R. Three-dimensional reconstruction and visualization of geological materials with IDL examples and source code[j]. Computers & Geosciences, 2001,27(4):419-426.
New Technologies for Engineering Research and Design in Industry 10.4028/www.scientific.net/AMR.971-973 Discussion on Development of Mineral Identification System Based on IDL 10.4028/www.scientific.net/AMR.971-973.1607 DOI References [2] Van Der Meer F, Bakker W. CCSM: Cross correlogram spectral matching[j]. International Journal of Remote Sensing, 1997, 18(5): 1197-1201. 10.1080/014311697218674