LUMS Home

Department of Computer Science > PhD Student > Arif Mahmood


Arif Mahmood

Area of Interest : Image registration, pattern search algorithms, video compression algorithms
Thesis Title : Fast Image Correlation over Large Search Space
Advisor : Dr. Sohaib A. Khan
Email : arifm AT lums DOT edu DOT pk

Personal Web Page: http://suraj.lums.edu.pk/~arifm/


Bio-Sketch

Achievements
Obtained First Position in MS Computer Science in LUMS
Received NMS Gold Medal in convocation 2003.
Remained on Dean’s Honor List, consistently during MS and during PhD in LUMS.
Obtained LUMS merit scholarship for PhD.
Obtained MoST scholarship for MS from ministry of Science and Technology.  
 
Education and Qualifications 
PhD Computer Science ( 2003-2008 expected) from  Lahore University of Management Sciences
MS Computer Science (2001-2003) from Lahore University of Management Sciences 
BE Engineering (1989-1994) from University of Engineering and Technology Lahore  
 
Industry Certifications: 
MCSE: Microsoft Certified System Engineer, 
OCP-DBA: Oracle Certified Professional Database Administrator,  
OCP-IAD: Oracle Certified Professional Internet Application Developer, 
Infrared Integration Engineer, Optical Fiber Theory and Practice 

  


Research Interests

Mostly my research work is around fast methods to compute correlation coefficient. I have focused on elimination algorithms and discovered two new elimination techniques for speeding up the correlation coefficient computation.  One technique is based upon transitive bounds upon correlation coefficient and the second is based upon the monotonic growth pattern of correlation coefficient.

I have also worked on block motion estimation in video encoders using correlation-coefficient as the similarity measure. This work focused upon video coding with linear compensation. I found that the variance of residue after linear compensation is always less than the variance of simple residue.

I have also proposed an elimination strategy for the AdaBoost based detector, that used the non-maxima suppression process to reduce the computations. The proposed technique is generic and can be used for any detector having monotonic detector response.




Recent Publications of Arif Mahmood

  • Exploiting Local Auto-Correlation Function For Fast Video To Reference Image Alignment

    Mahmood, A., and Khan, S., In IEEE International Conference on Image Processing (ICIP 2008), San Diego, CA, USA, October 12-15, 2008.

  • Structure-Less Object Detection Using AdaBoost

    Mahmood, A., In International Conference on Machine Vision 2007, Islamabad, Pakistan, 2007.

  • Exploiting Inter-Frame Correlation for Fast Video to Reference Image Alignment
    Mahmood, A., and Khan, S.A., In Asian Conference on Computer Vision (ACCV 2007), Tokyo, Japan, November 2007.
  • Early Termination Algorithms for Correlation Coefficient Based Block Matching

    Mahmood, A., and Khan, S.A., In IEEE International Conference on Image Processing (ICIP 2007), San Antonio, TX, USA, September 2007.

  • Video Coding using Linear Compensation (VCLC)

    Mahmood, A., Uzmi, Z.A., and Khan, S.A., In International Conference on Communications, ICC 07, Glasgow, UK, June 24-27, 2007.

  • Automatically Generating Association Rules Under Diverse Operational Conditions for a Large Scale Power Plant

    Awais, M.M., Mahmood, A., and Karim, A., In 2nd International Bhurban Conference on Applied Sciences and Technology, Islamabad, Pakistan, June 2003.