B-4/7. M.S.FLATS, • PESHWA ROAD, • GOLE MARKET, • NEW DELHI-110001

PHONE (011) 3345996 • E-MAIL ojaswa_sharma@rediffmail.com

7^th January 1982

• Ranked top in the class throughout senior years of Schooling.

• Secured admission to the IIT Roorkee, through all India competition. The institute is the best in civil engineering education in Asia and has a separate building for remote sensing works.

                        Record at I.I.T:

• Always remained among the top students and have always got full merit scholarship during my undergraduate study.

• My current GPA is 7.89 / 10.00 in a class of 68.

1997                 AISSE (CBSE)                                              New Delhi

All India Senior Secondary Examination

 §     Obtained certificate of merit from CBSE from being in top 0.1% of the student in Mathematics

§         Obtained 77.6% aggregate marks [mathematics 97%, Science 97%]

1999                 AISSCE (CBSE)                                           New Delhi

All India Senior Secondary Certificate Examination

§         Obtained 84% aggregate marks [Mathematics 97%, Physics 95%]

• Scholarship and Certificate of merit awarded by the

            Central Board of Secondary Education, for being among

            The top 0.1% of the successful candidates appeared in

            The Delhi Board Secondary School Examination, held in

            1997.

• Always got merit scholarship in the three years of  undergraduate study.

• Got third certificate in a quiz competition organized by the

      The Indian Society of construction materials and   structures.

• Got second prize in annual exhibition for 3D modeling using computer graphics in C.

• Presented a paper on the use of Fractals in linear feature   extraction from satellite imagery and won I prize in the National level paper presentation competition organized at national level.

• Also presented a paper on the art of Steganography for data hiding by integrating a security technique based on random numbers and implemented it through a GUI application for 24 bit images.

            Got Ist prize at BITS pilani for this.

• Developed a piece of software (GUI) for generating colored FRACTALS (IFS) by interactively choosing the affine transformations, running it and saving it.

• Special interest in computer graphics. Had a course in Computer Graphics. Have knowledge of 3D Computer Graphics and some experience in graphics in C. Have studied Graphics programming using OpenGL.

• Have great interest and passion for FRACTAL science. Have studied Fundamentals of fractal science, conventional                                                                                                                                                                              fractals like Mandelbrot set, Julia set, Seirpinski Gasket, etc., Iterated Function Systems (IFS).

• Have studied the Win32 API programming and using it in             Visual Basic
• Have interest in Remote sensing and Digital Image Processing and have also done projects in this field.[Details attached]

 §        Volunteered in the annual Paper presentation competition and coordinated the software quiz event in my junior year.

 §      Remained an active member of SDG (Software Development Group)

 §         I am conversant with QBasic, C, C++ and Visual Basic.

        §         Have worked on UNIX (X-Windows), LINUX, DOS, and Windows (98,2000 & XP) based machines.

 § I am familiar with software packages like MATLAB, MathCad GRASS (GIS), STAAD etc..

 §    Knowledge of Win32 API's and OpenGL.

Hindi (Mother Tongue)

English (Proficient)

Have done two month’s Practical Training at ATREE (Ashoka trust for Research in Ecology and the Environment), Bangalore, INDIA.

Project on fractals

TOPIC             :Development of an interactive application for the generation of fractals

Advisors: prof. R.C. mittal ( Deptt. of mathematics)

Duration: May 2001 - july 2001

            This project was an attempt to develop a highly interactive application, which generates a kind of fractal objects through recursion. Here, the fractal objects known as IFS (Iterated function systems) were generated taking graphical inputs in the form of affine Transformations (rotation, translation, scaling, shear, reflection) and then applying these to create the fractal.

The application was developed in Visual basic and had features of creating colored IFS and saving them.

II year project

TOPIC  :  Linear feature extraction from satellite imagery using Fractals

Advisors:      Prof. Manoj Arora

Duration:      June 2001 - May 2002

            After reading a lot about fractals and IFS (iterated function systems), i tried to apply the concept of fractal dimension to satellite imagery. as per the Euclidian geometry, the dimension of an object can only be 0,1,2,or 3 ( ...and of course 4 if we consider the time frame too.), but fractal science goes beyond it. It says that point objects if grow very unwildly, approach towards linear curves and tend to attain a dimension of 1.similarly, if linear one dimensional curves become more and more complex, their dimension tends towards 2.this is true for other dimensional objects too.

            Thus, a fractal curve or object can have fractional dimension depending on its complexity. Fractals are objects of ultra high symmetry and repeatedness. They are iterative in nature and have no effect of zooming. Nature exhibits symmetry and hence natural images can safely be taken to match with fractals to a certain extent. My research work uses the fact that fractal dimension of linear objects lies between 1.0 and 2.0.   Therefore linear features existing within an image can be separated by taking out all the features having dimension between 0 and 1.

            There are several techniques to calculate fractal dimension of images. Here the technique used is the local fractal dimension, which is calculated on pixel basis. This was implemented in MATHCAD.

The   results, which have been obtained so far, are highly promising.

III year project (...contd.)

TOPIC   :    Development of a fractal based image classification algorithm.

            After successfully applying fractal dimension for feature extraction, i tried to use the same for image classification. The concept remains the same while the methodology changes. Here, i tried to clump areas of similar fractal dimension and assign them a different class.

            I am currently trying to incorporate some textural properties to it so that the classification becomes more effective. I implemented the concept in MATLAB; the initial results required more refinement.

Summer Project  1

TOPIC : Feature extraction from scanned toposheets and auto   digitization.

Guide    :     Dr. Jagdish Krishnaswamy

Duration:     May 2002 - July 2002

            This project was a part of the summer training which i did at ATREE, Bangalore. In a country like India, where digital thematic maps are not freely available, such a mechanism of feature extraction is very useful. Usually the manual digitization takes months to complete. My works suggests a technique which helps in the extraction of contours and streams out of a scanned toposheet.

The process starts with separating the R, G, and B components of a 24-bit image.

            These are now processed digitally to remove apparent noise using some filters. I used fast Fourier transform to remove the noise. Now the   8-bit images thus obtained are taken and sliced into 8 binary planes (a byte = 8 bits).what i observed was that all the bit planes contained different amounts of information. And there are usually two bit planes in an 8-bit image which contain very precise information. Like contours or streams or both.

So after choosing the most relevant binary images these are processed again to remove any noise if they contained. Since i was also learning GRASS GIS at that time, i tried doing this all in GRASS. The raster images thus obtained were very good. Then it was a question of converting these raster file to vector and GRASS really helped in it.

            Both contours and streams were vectorised.

The only problem was with contours. They were discontinuous and disjoint. Initially i felt that some curve reconstruction method can help, but that was difficult to implement at that stage so i used the GRASS’s build and clean operations on the vector file. This solved the problem to a large extent and with a small manual work things can work right. After a DEM is created labels can be attached easily and a DEM can be easily prepared.

The last two tasks can be easily performed in GRASS. This technique is in its primitive stage and i am sure, with a little improvement, it can provide a very efficient and faster way to digitize maps. This can be cost effective also if implemented through GRASS (since it is freely available)

Summer Project 2

TOPIC    :  Learning GRASS GIS and module development

Guide     :  Dr. Jagdish Krishnaswamy

Duration :  May 2002 - July 2002

            In the same NGO, i learned to work in GRASS environment, which is an open source GIS.It is a Linux based GIS and is developed in C language.

After going through the functioning of GRASS, i also learned how to develop modules for GRASS. Compiling them and integrating with GRASS.

For the two ongoing projects, i made two GRASS modules which helped in the projects. These are:

            Module 1:

            To compare the performance of maximum likelihood classifier with that of a new non-linear rule based classifier ( Krishnaswamy J., TREE model). Here i looked into the already existing implementation of the maximum likelihood classifier. This actually calculates probability of each and every class compares them and stores the maximum probability and prepares a classified map. I then developed a separate module which collects the thrown out probabilities and makes probability maps out of these.

            Module 2:

            This module was an integration of heterogeneity parameters of a raster map. For this one of the parameter required was the shape factor. Now shape factor is also an indicative of the complexity of the object. So for this, we all decided to have fractal dimension as the desired parameter. For this i developed a module in GRASS which calculates the localized fractal dimension of the raster maps provided with two suitable kernel sizes.The output is a fractal dimension map (floating type)

Later on an index was derived from such a map as the required parameter.

Paper 1

Topic       :      Linear feature Extraction using Fractals 

Advisors  :      prof. Manoj Arora

authors    :      Sharma O., Shah N.         

Time        :      January 2002

Place       :      IIT -Roorkee

        As part of my home paper, I did an extensive study on the applicability of fractal dimension on natural images. The paper shows an example of extraction of river (Ganga canal) and roads from a LISS image.

The paper was very much appraised and got first prize out of a total of 96 papers from all over India.

paper 2

Topic                :           Steganography,new frontiers in data hiding

authors              :           Sharma O., Shah N.      

Time                 :           March 2002

Place                 :           BITS-Pilani (Birla Institute of                                                       technology, Pilani, Rajasthan)

            This paper actually was a result of our study on a new technique of data hiding. This technique, known as Steganography actually hides any text inside images. This harnesses the fact that images generally have no information in the least significant bit of their 8-bit color value. Using this fact together with a new idea of randomness, we were able to develop a full fledged application for data hiding inside 24-bit bitmap images. The extraction of the text (or rather decoding) can only be done from the decoder provided in the application. Even if one gets the images, these will be of no use to him. Furthermore, the corrupted images look normal.

This paper got the first prize.