Nurul Wahidah Arshad

Digital watermarking for recovering attack areas of medical images using spiral numbering

Medical image is seen as one of crucial data that demand for authentication method in order to ensure that data has not been tampered. This paper has conducted a research proposing a recovery feature in an authentication watermark with spiral manner numbering, which shows a good numbering system of embedding. The schema is a fragile, blind watermarking. The operating time taken is short and the result is very promising with high Peak-Signal-to-Noise-Ratio (PSNR) value and low Mean-Squared Error (MSE).

Speech processing for makhraj recognition

In our daily day, improvement of makhraj for Arabic alphabets is a topic that very useful in many applications and environments. The existing system cannot recognize the appropriate pronunciation of each alphabet with the existence of noise. As an example “ha”, with the disturbance from the noise, the system may recognize wrong alphabet like “kho”. This paper focus on noise removal in makhraj recognition using Least Mean Square (LMS) Algorithm based on Adaptive Filter to search for the optimal solution to adaptive filter, including system identification and noise cancellation. There are 30 Arabic alphabets from until. However, this project will only use 7 alphabets as samples that are from until. The speech processing will be used to obtain same waveform output from two different situations. The filtered data will be processed to match the standard pronunciations and it will be integrated with filter design process in MATLAB. As a result, the waveform of noise cancellation using LMS algorithm is quite similar with the waveform of reference signal. As a conclusion, it is proved that noise cancellation method remove noise from unknown system.

Unclean hand detection machine using vision sensor

This paper will discuss about the unclean hand detection using vision sensor for an automated hand wash screening system. Currently the hand wash screening audit is done manually by an expert to monitor the hand under ultraviolet light once its been washed. Hence, there is a need for more human experts to conduct the screening manually. This project is proposed to automate the hand wash screening audit by using a vision system. The vision system is designed to increase accuracy to detect the unclean area of washed hands. This system will not only detect the unclean area, but will also calculate the percentage of the unclean area which will be used as further analysis of the efficiency of the system. However, we need to build the hand wash prototype using ultraviolet light and a camera that is connected to the computer to process and display the results of hand wash screening.

Feature extraction of pus cells detection and counting in sputum slide images

This paper discusses on a feature extraction of pus cell in sputum slide image. This invention is developed to analyze and count the content of cells specifically pus cell within a biological sample, and more particularly sputum sample which is useful for sputum quality grading. This pus cell detection is addressed by mean intensity and area for single and overlapping pus cells. It is found that mean intensity and area for single pus cell are ranges from 130 - 163 and 35 - 81 respectively. Whereas, with considering other elements exist in sputum image such as epithelial cells and artifacts, the mean intensity for overlapping pus cells are ranges from 130 - 45 with area of 65 - 300. This system reliability is above 80% from the validation results.

Speech processing for makhraj recognition: The design of adaptive filter for noise canceller

In our daily life, improvement of makhraj for Arabic alphabets is a topic that very useful in many applications and environments. The existing systems cannot recognize the appropriate pronunciation of each alphabet with the existence of noise. As an example of “ha”, the system may recognize wrong alphabet like “kho”. This paper focuses on noise removal in makhraj recognition using Normalized Least Mean Square (NLMS) Algorithm based on Adaptive Filter to search for the optimal solution. There are 30 Arabic alphabets from **** until ****. However, this project will only use 7 alphabets as samples, they are **** to ****. The speech processing is used to obtain same waveform output from two different situations. The filtered data is processed to match the standard pronunciations and it is integrated with filter design process in MATLAB. From the result, the waveform of noise cancellation using NLMS algorithm is quite similar with the waveform of reference signal. It is proved that noise cancellation method remove noise from unknown system.

A Comparison of Model Validation Techniques for Audio-Visual Speech Recognition

This paper implements and compares the performance of a number of techniques proposed for improving the accuracy of Automatic Speech Recognition (ASR) systems. As ASR that uses only speech can be contaminated by environmental noise, in some applications it may improve performance to employ Audio-Visual Speech Recognition (AVSR), in which recognition uses both audio information and mouth movements obtained from a video recording of the speaker’s face region. In this paper, model validation techniques, namely the holdout method, leave-one-out cross validation and bootstrap validation, are implemented to validate the performance of an AVSR system as well as to provide a comparison of the performance of the validation techniques themselves. A new speech data corpus is used, namely the Loughborough University Audio-Visual (LUNA-V) dataset that contains 10 speakers with five sets of samples uttered by each speaker. The database is divided into training and testing sets and processed in manners suitable for the validation techniques under investigation. The performance is evaluated using a range of different signal-to-noise ratio values using a variety of noise types obtained from the NOISEX-92 dataset.

HILBERT-LSB-C as authentication system for color medical images

This paper proposes a new numbering method for a fragile watermarking algorithm aimed at improving color medical image watermarking. The proposed method uses Hilbert pattern numbering before watermarking operations such as parity bits check and comparison between average intensities as the authentication data. The authentication data embedded in the same host image are utilized to localize any tamper using block-wise approach. The method is very effective since it only requires a secret key and public, chaotic mixing algorithm to recover the attacked image. We use the Hilbert mapping approach, which is more compatible with medical image modalities, which is not only specifically to the square shape of image but applicable to all kinds and modalities of the image. We propose the algorithm to match the criterion of having 3 planes in a color image. The peak-signal-noise-ratio value of the proposed scheme is very good, achieving up to 56 decibelf.

Authentication system for medical images using Hilbert numbering

Medical image is seen as one of crucial data that demand for authentication method as it is highly confidential and used in insurance claim, evidence of jurisdiction and personal identification. Nowadays, Hospital Information System (HIS) is used widely at hospitals and clinical departments and it handles thousands of crucial electronic data in medical. We have introduced a fragile watermarking method using spiral manner numbering which showed a good numbering system and excellent embedding, but due to the technique, it only embedded in square shape. We enhanced the scheme to the Hilbert numbering scheme, which is more compatible with medical image modalities, which is not only specific to square shape of image but applicable to all kinds of image.

The detection and summation of squamous epithelial cells for sputum quality testing

Sputum with good quality is important to detect diseases. The quality of sputum is determined using Bartletts Criteria by considering the score of squamous epithelial cells (SEC), pus cell (neutrophils) and macroscopy. For squamous epithelial cells, the score is 0 if SEC is less than 10. Whereas if SEC is within 10 to 25, the score is -1 and the score is -2 if the number of SEC is greater than 25. Currently, the detection of SEC in sputum is manually done by technologists. However, the problem with manual detection is time consuming. So, an automated vision system using image processing technique to detect the sputum quality is desirable. Image processing such as image segmentation is used to detect and count the number of SEC, and then the score is determined. Lastly, the percentage of error for this project is calculated.

Automated detection and counting of pus cells on sputum images

This paper discusses on a feature extraction of pus cell in sputum slide image. This invention is developed to analyze and count the content of cells specifically pus cell within a biological sample, and more particularly sputum sample which is useful for sputum quality grading. This pus cell detection is addressed by mean intensity and area for single and overlapping pus cells. It is found that mean intensity and area for single pus cell are ranges from 130-163 and 35-81 respectively. Whereas, with considering other elements exist in sputum image such as epithelial cells and artifacts, the mean intensity for overlapping pus cells are ranges from 130-45 with area of 65-300. This system has the sensitivity of 92.11%, with specificity of 81.82%, accuracy of 87.32% and precision of 85.37.