International Journal of Computer Engineering and Applications, Volume XI, Issue XII, Dec. 17, ISSN

Transcription

1 AUTOMATIC EXTRACTION OF PROFILE FROM AN IONOGRAM USING DIGITAL IMAGE PROCESSING Bitap Raj Kalita 1, 2, Sankar Jyoti Nath 1, P.K.bhuyan 1, Ajay Khandare 3 and Anil Kulkarni 3 1 Centre for Atmospheric Studies 2 Department of Physics Dibrugarh University Dibrugarh, India 3 SAMEER, Mumbai Corresponding author:bitapkalita@gmail.com ABSTRACT: An Ionogram recorded by sweep frequency HF RADAR provide vital information of the ionosphere for HF band communication systems as well as atmospheric research community. For real or near real time use of this information by users, the ionograms need to be scaled easily and quickly. A digital image professing based autoscaling software was developed for this purpose and tested on hundreds of ionograms. The algorithm, working of the software and the test results are presented in this report. The software is shown to be robust, easy to use and relatively accurate. The software is a generalized package as it takes only an image as input, and therefore, can work with ionograms recorded by any commercial or in-house ionosonde. Such a hardware independent package should be extremely handy for the user community. Keywords: Ionosonde, Ionogram, Digital Image processing, HF RADAR [1] INTRODUCTION Ionosphere is an upper atmospheric layer of weakly charged plasma. It can influence RF e-m wave propagation, and, therefore it is studied using Sweep Frequency (HF band) RADAR called Ionosonde. The ionosonde provides the sounding frequency and corresponding height of vertical echo in the form of an ionogram. Actual daytime and nighttime ionogram recorded by a commercial ionosonde is given in Fig 1 which shows the E and F layers of the ionosphere. The two half parabolic traces represent the echoes of the ordinary wave and extra-ordinary wave from the F layer of the ionosphere as the ionosphere is anisotropic in presence of the earth s magnetic field. The ionogram needs to be scaled, often manually to extract parameters useful for estimating HF band communication channels. The manual scaling of the ionogram is a very tedious, time consuming and subjective process. Different Bitap Raj Kalita, Sankar Jyoti Nath, P.K.bhuyan, Ajay Khandare and Anil Kulkarni 9

2 AUTOMATIC EXTRACTION OF PROFILE FROM AN IONOGRAM USING DIGITAL IMAGE PROCESSING types of ionosondes are commercially available, and, in house proprietary ionosondes are also in operation. These produce different quality ionograms each with different data formats. Software have been developed to automate the scaling process for the popular digital ionosondes like Digisoinde [1] called ARTIST. and Canadian Advanced Digital Ionosonde. These are based on data recorded by the ionosonde. Other autoscaling software like Autoscala and Cadiscala [2] were developed using image recognition technique. None of these are universal in the sense that the software work well for the target ionosonde only. The need for the user community are ionospheric parameters like critical frequency, real height and MUF etc which should be easily scalable from any type of ionogram. The task of scaling or auto scaling different ionogram with one [2] METHODOLOGY The ionograms are images that contain all the information retrieved by the RADAR for one cycle of frequency sweep. Therefore, to scale the parameters like critical frequency, height or MUF the image is sufficient and under lying data is not important. Therefore, we have used the images of the ionograms recorded by different ionosondes like CADI, Australian IPS 4 and other ionograms images available in the internet. These are converted to greyscale digital images and provided as input to the software module. The software is based on image processing principle which is described in detail below in section 2.1. The ionogram is then processed for real height [3] by using POLAN [4]. [3] DIGITAL IMAGE PROCESSING Fig1: An ideal ionogram Digital image is composed of finite number of elements, each of which has particular location and values [5]. These elements are referred to as picture elements, image elements or pixel, pixel is widely use to denote the elements of digital image processing, basically a digital image is nothing but only a collection of zeros and ones. The digital image is divided into pixels. Let us consider m number of pixels along with horizontal and n along the vertical direction. The image hence resolution of m x n and the image is stored as m x n matrix, with each element being the intensity value of that pixel. For a greyscale images, the pixel value is a single number that represents the brightness of the pixel. The most common pixel format is the byte image, where this number is stored as an 8-bit integer giving a range of possible values from 0 to 255. Typically zero is taken to be black, and 255 is taken to be white. The intensity scale is shown in Fig2 Fig2: Intensity Scale of greyscale image Bitap Raj Kalita, Sankar Jyoti Nath, P.K.bhuyan, Ajay Khandare and Anil Kulkarni 10

3 For a colour image there are three or four (RGB or CMYK) sets of intensities corresponding to each colour component. The intensity scale of RGB image shown in Fig3. Matlab has an image processing tool with some useful features that we can use in various applications. Fig3: Intensity scale of RGB colour [4] THE ALGORITHM The program converts the input image into greyscale and extract the points with respect to the given intensity range. For example if we have a black line plot with a white background, the points containing (or comprising) the line have an intensity of 0 and the background has an intensity of 255. So we can get all the points containing the line by simply scanning for zero intensity points. However, when we convert a colour line plot with white background to gray-scale, we have intensities below 255 for the line but not necessarily zero. So we can scan between an upper and lower intensity limit to digitize the line plot. The flow chart for the image processing program is shown below in Fig 4 and Fig 5. Stored.jpg in a folder Count file size N =0 Display out the digitized points n<n Save output data as.txt Call the profile extraction algorithm STOP n= n+1 Fig4: Flow chart of the GUI Bitap Raj Kalita, Sankar Jyoti Nath, P.K.bhuyan, Ajay Khandare and Anil Kulkarni 11

4 AUTOMATIC EXTRACTION OF PROFILE FROM AN IONOGRAM USING DIGITAL IMAGE PROCESSING Input as.jpg file Convert the image file into grey scale Set maximum and minimum value of X axis and Y axis Set the intensity range between 0 to 255 Scan for the low intensity points and digitize it Measure the position of the digitized points with respect to X and Y axis Extract the value of X and Y axis Save output as.txt Fig5: Flow chart of the algorithm Bitap Raj Kalita, Sankar Jyoti Nath, P.K.bhuyan, Ajay Khandare and Anil Kulkarni 12

5 Fig6: Screenshot of ionograms as.jpg file [5] IMPLEMENTATION Fig7: Digitized points extracted from the ionogram. A Matlab program to implement the algorithm was developed along with a graphical user interface. The Matlab program was used to automatically extract whole day ionogram profiles in a single click. The images of the ionogram are kept aside in a directory with the respective date. The GUI includes one push button one list box and one axes. After Clicking the push button (START button) all images will display in the list box and each of images will go through the profile extraction algorithm and output file of each ionograms is stored in the given address. [6] RESULTS Fig 6 shows the image used as input to the software and Fig 7 shows the resultant filtered and extracted data point. With the help POLAN program the real height analysis of ionograms is carried out. The real height analysis is shown below in Fig 8. Bitap Raj Kalita, Sankar Jyoti Nath, P.K.bhuyan, Ajay Khandare and Anil Kulkarni 13

6 AUTOMATIC EXTRACTION OF PROFILE FROM AN IONOGRAM USING DIGITAL IMAGE PROCESSING Fig8: Real height profile estimated from ionogram. The coloured trace is the ionograms and the black dots are the retrieved real heights. [7] PERFORMANCE ANALYSIS The program is written in Matlab7.9.0 (R2009b); it runs under UBUNTU, Windows8,7, 98, and XP. Minimum system requirements: 133MHz processor, at least 500Mb of RAM, SVGA graphics of 1024_768 pixels. System should have Matlab installed in it. There is no requirement to install the implemented software since it is a standalone.exe file. The Ionogram profile extractor module has been tested on 1000 ionograms with 15 min interval in a single click. The Ionogram profile extractor module has been tested on 1000 ionograms with 15 min interval with a success rate of 80% [8] DISCUSSION AND CONCLUSION The software module is developed to extract ionospheric information from already available digital image in form of ionograms using digital image processing toolkit of Matlab. The algorithm and the results are provided. The performance is briefly studied and is observed to be sufficiently good but requires further improvement for use in real time and in commercial application. The ability of the software to process data from different instrument seamlessly is considered its primal advantage. ACKNOWLEDGEMENT The research was funded by the Department of Electronics and Information technology (DEIT), Government of INDIA. Then work was carried under the collaboration project between SAMEER and Dibrugarh University. Sankar Jyoti Nath is grateful to DEIT and SAMEER, Mumbai for the fellowship under this project. Bitap Raj Kalita, Sankar Jyoti Nath, P.K.bhuyan, Ajay Khandare and Anil Kulkarni 14

7 REFERENCES [1] Reinisch, B. W., et al. New Digisonde for research and monitoring applications, Radio Sci., 44, RS0A24, doi: /2008rs004115, [2] Pezzopane, M., Scotto, C., Highlighting the F2 trace on an ionogram to improve Autoscala performance, Computers & Geosciences 36, ,2010 [3] Liu,Y.J, F.T Berkey, S.L. Wu, TAO, Vol.3, No.2, , June A Study of True Height Analysis Methods, [4] Titheridge J,Ionogram analysis with the generalized program POLAN, World Data Center Report, UAG-93, WDC-A-STP, Boulder, CO, 194,1985. [5] Rafael C. Gonzalez and Richard Eugene Woods, Digital Image Processing (2nd Edition)Prentice Hall,2002 Bitap Raj Kalita, Sankar Jyoti Nath, P.K.bhuyan, Ajay Khandare and Anil Kulkarni 15