Muhammad Sirajul Islam

Research Assistant

Qualifications
Phone +61 (3) 9905 FAX +61 (3) 9905 3637

Email:Muhammad.Islam@sci.monash.edu.au

RESEARCH INTERESTS

Muhammad Sirajul Islam (Siraj) is investigating novel techniques to remove distortion, blur (by estimating PSF) which is inherent in many image capturing techniques including X-ray imaging. He is also investigating applicability of various sophisticated algorithms in modern graphics processing unit (GPU) to increase computational performance. These will improve the lung volume analysis from phase contrast images which is a continuing research project developed by Dr Marcus Kitchen collaborated with researchers from the Monash Department of Physiology.

Siraj's research interests include computer vision, image processing, pattern recognition, medical imaging and general purpose computations on GPU (GPGPU). Visual grouping is of paramount importance and a challenging contemporary research topic in image processing and computer vision, embracing many disparate application domains, ranging from image recognition and object detection through to medical diagnoses, vision based robot navigation and industrial automation. In many of these application areas, precise automatic object segmentation requires higher-level perceptual features that are constructed from lower-level primitives. However, a number of factors such as noise and distortion in the lens of the camera affect the accurate extraction of these visual primitives. Despite considerable endeavour, improving vision processes to efficiently deal with noise, image distortion, achieving good accuracy and fast processing capability still remains a major goal in computer vision technology. Recently growth in the computational power of the graphics hardware and its flexible programmability for general purpose computations offer great possibility for the acceleration of computer vision algorithms. Siraj has already developed new methods to perform visual grouping on distorted images directly. The traditional approach performs visual processing on distorted images by dewarping and then applying image processing and computer vision operation on the dewarped image. However, in this approach operations are directly applied on a distorted image by taking appropriate action to handle the distortion effect. He has also developed novel technique to perform Connected Component Labeling (CCL) on modern graphics processing unit (GPU).

SELECTED PUBLICATIONS

"Nonlinear Similarity Based Image Matching", Islam, M.S., Kitchen L., 2006, in IFIP International Federation for Information Processing, Volume 228, Intelligent Information Processing III, eds. Z. Shi, Shimohara K., Feng D., (Boston: Springer), pp. 401-410.

"Gradient Correction of Distorted Image: A Differential Chain-rule Approach". Muhammad Sirajul Islam and Les Kitchen Submitted to Pattern Recognition Letter, Elsevier.

"GPU based 1D Run Labeling: A New Approach". Muhammad Sirajul Islam and Les Kitchen, In preparation to submit in a Conference.

"Efficient Connected Component Labeling on Programmable Graphics Processing Unit (GPU)", Muhammad Sirajul Islam and Les Kitchen, In Preparation for submission in a Journal.