Projects by Dr. Ghulam Muhammad

Funded: (Six funded projects)

A. 2011

Non-intrusive Image Forgery Detection using Multiresolution Framework

Funding Body

KSU-KACST National Plan for Science and Technology (NPST).

Abstract

In this project, we will develop a system for reliable and robust non-intrusive forgery detection using multiresolution framework based on multiresolution framework and statistical modeling. The outcome of the project will be useful in helping the Saudi security agencies and commercial organizations such as insurance companies to establish the authenticity of pictorial information and to validate the evidence presented in the form of image data in front of court.

Duration: Two years

Responsibility: Principal Investigator (PI) with co-PI: Dr. Muhammad Hussain, Prof. George Bebis, and Prof. Anwar Mirza

Amount: 1.618 million Saudi Arabian Riyals

B. 2010

Category Specific Face Recognition

Funding Body

KSU-KACST National Plan for Science and Technology (NPST).

Abstract

This project aspires to advance the state of art in face recognition by investigating a novel approach to face recognition using category-specific recognition. There are two key ideas behind the proposed approach: first, classifying faces into different face categories using information from various visual cues (e.g., gender, ethnicity, age, etc.); and second, designing “category specific” or “specialized” recognition processes (e.g., Caucasian male, between 20 and 30 years old) by exploiting the most discriminatory features within each face category.

Duration: Two years

Responsibility: Co Investigator with PI: Dr. Muhammad Hussain and Co-I: Prof. George Bebis and Prof. Anwar Mirza

Amount: 1.367 million Saudi Arabian Riyals

C. 2010

Arabic Speaker Recognition

Funding Body

KSU-KACST National Plan for Science and Technology (NPST).

Abstract

In this project, we are tackling speaker recognition aiming to contribute to the existence of a robust, channel independent, Arabic Speaker Recognition system. The main objective of this system is to add some new values to the Arabic Database recognition, as well as participating in the establishment of some new techniques and standards in the Kingdom of Saudi Arabia. We are intending to use some state of the art techniques, mainly the GMM-UBM generative modelling techniques, as well as the SVM classification. The approach of combining both GMM-UBM/SVM will be a challenge to be introduced in the Arabic speaker recognition technology.

Duration: Two years

Responsibility: Co Investigator with PI: Dr. Mansour Alsulaiman

Amount: 1.098 million Saudi Arabian Riyals

D. 2009

Environment Detection for Digital Audio Forensic

Funding Body

Center of Excellence for Information Assurance (CoEIA), King Saud University.

Abstract

In our proposed project, we focus on detecting the environment from the recorded speech for forensic applications. We discover in this project: (1) is it possible to detect environment from audio stream; (2) if yes, with how much accuracy; and (3) how to increase that accuracy.

Duration: Six months (February – August)

Responsibility: Principal Investigator

Amount: 60000 Saudi Arabian Riyals

E. 2008-2009

Pitch Detection in Noisy Environments

Funding Body

CCIS Research center, King Saud University.

Abstract

The objective of the proposed research is to develop a noise-robust, real time PDA, which is easy to integrate with other applications. AMDF based algorithms are suitable for real time operations, but suffer from incorrect pitch detection in noisy conditions. The proposed extended AMDF (EAMDF) involves in sufficient number of averaging for all lag values compared to the original AMDF, and thereby eliminates the falling tendency of the AMDF without emphasizing pitch harmonics at higher lags, which is a severe limitation of other existing improvements of the AMDF.

Duration: One year

Responsibility: Principal Investigator

Amount: 36400 Saudi Arabian Riyals

F. 2007-2009

Study on canonicalization for robust speech recognition

Funding Body

JSPS Post-Doctoral Research Fellowship, Japan. (The Ministry of Education, Culture and Sports, Japan)

Abstract

Canonicalization is a process for eliminating the influence of various hidden factors in automatic speech recognition (ASR) that decrease acoustic likelihood between a targeting category and the other categories. The hidden factors include acoustic transfer function, gender, noise types and their SN ratios, age (child/old), speaking style, phone context, and so on. The proposed canonicalization process maps an input acoustic feature onto a DPF space, which is invariant to acoustic environments, widens the margin of the acoustic likelihood, which results in less number of mixtures in HMM, and consequently gives a functionality of a high performance phonetic typewriter. To realize the canonicalization process, we simulate a human auditory system at the upper level of the front-end in our ASR system. The goal of the research was to solve the problem of coarticulation of phonemes at their transitions and thereby increase recognition performance.

Duration: Two year (Shortened for joining KSU, Riyadh)

Responsibility: Principal Investigator

Master’s Thesis and Project Supervision

2010-2011

Applying Feature Selection on Local Binary Patterns/WLD for Ethnicity classification for Category-specific Face Recognition

Student: Fatimah Alanizi, [Project]

2009-2010

Feature selection based verification system using palm and fingerprint

Student: Eng. Muhanad M. Jazzar, [Thesis]

2009-2010

Extract context from environment sound

Student: Mobarak Obaid Alqahtani, [Project]

Student Project: undergraduate

2010-2011

Implementation of Arabic speaker recognition system

Student: Mohamed Almalki and Abdul Aziz

2010-2011

Implementation of automatic diagnostic system for medically disordered voice

Student: Mohammed Alomari and Mohammed Alshehri

2009-2010

Implementation of image forgery detection

Student: Khalid Khawaji

2009-2010

Effect of vehicle noise in Arabic speech recognition

Student: Majed Khaled Al Hamsh

2009-2010

Environment detection from digital audio

Student: Waleed Al-shaya

2009-2010

Study of Pitch Pattern in Saudi Accented Arabic Speech

Student: Mohammed Salih Al-Otaibi

2008-2009

Evaluation of Speech Enhancement Algorithms

Student: Mohannad Samir Khattab

2008-2009

Removing Noise from Noisy Speech

Students: Khaled Abdullah Al-Otaibi and Hussain Hilal Al-Qurashi