منصة إيفاد العلمية

همسة ضياء مجيد

إرسال رسالة

التخصص: الهندسة الإلكترونية

الجامعة: الجامعة التكنولوجية

النقاط:

معامل الإنتاج البحثي

الخبرات العلمية

مهندسة - استاذة جامعية
استاذة جامعية

الأبحاث المنشورة

A System Model Based on Slantlet Transform to Estimate Optical Flow

المجلة: IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING

سنة النشر: 2013

تاريخ النشر: 2013-07-22

The estimation of optical flow is the basic step for many engineering applications that exploit the image processing field as a part of their models. In this paper a model called Slantlet Based Optical Flow Estimation (SLT_OFE) is proposed to estimate the optical flow. Slantlet Transform (SLT) used as an effective tool, 2D and 3D- SLT- Level 2 (SLT2) are computed and employed in the proposed model to provide high accuracy estimation of the optical flow. By its definition, optical flow is a velocity field, so the velocities in this paper are computed using the widely used Differential Technique. Two methods from this technique are adopted; Horn-Schunck Method and Lucas-Kanade Method. The optical flow is estimated for two types of image sequences; synthetic sequences and real sequences. Unlike the real sequences, the synthetic sequences have known true velocities which are used for evaluating the proposed model by calculating mean error (Mean Err.), angular error (Ang. Err.) and Standard Deviation (STD). For extreme study of its performance, the proposed SLT_OFE is compared with three other models that are based on level two (Discrete Wavelet Transform (DWT), Discrete Multi Wavelet Transform (DMWT) and Framelet Transform (FT) ) which are implemented in this paper and employed in the conventional models, 2D-DWT2_OFE, 2D-DMWT2_OFE and 2D-FT2_OFE. The results show that the proposed model offers minimum values in errors and STD when 2D-SLT2 is used, and these results are improved by using 3D-SLT2. This leads to the fact that the proposed model SLT_OFE through both of 2D and 3D approaches possesses an improvement in the optical flow estimation process with higher accuracy, than the other models produced in the same circumstances. MATLAB Version 7.12 (R2011a) is used to implement the proposed model and the conventional models.

Assessment of sustainability indicators for green building manufacturing using fuzzy multi-criteria decision making approach

المجلة: Journal of Cleaner Production

سنة النشر: 2020

تاريخ النشر: 2020-07-15

Over the past decades, building manufacturing has caused serious environmental impacts, despite its role in the national economic growth. Thus, in developing strategic plans for economic growth, many governments consider the application of green manufacturing building and technologies as key factors towards a greener economy and lower carbon emission. However, so far, there have been limited efforts relating to the application of eco-efficiency ideas in building manufacturing. In fact, environmental sustainability in building project and delivery is still at a nascent stage. Thus, this study aims to identify and rank the sustainability indicators for assessing green building manufacturing in Malaysia by considering Green Building Index (GBI), which is the most applied sustainability rating tool in the country. Data is collected from a panel of experts and fuzzy Decision-Making Trial and Evaluation Laboratory (DEMATEL) is performed to reveal the importance level and relationships among sustainability indicators in green building manufacturing. Results show that “Energy Efficiency” and “Indoor Environmental Quality” are the most important, while “Water Efficiency” and “Innovation” are the least important criteria in assessing green building manufacturing in Malaysia. This study can serve as a guideline to select and promote the optimum practices in green building manufacturing.

Text Detection on Images using Region-based Convolutional Neural Network

المجلة: UHD Journal of Science and Technology

سنة النشر: 2020

تاريخ النشر: 2020-08-02

In this paper, a new text detection algorithm that accurately locates picture text with complex backgrounds in natural images is applied. The approach is based primarily on the region-based convolutional neural network anchor system, which takes into account the unique features of the text area, compares it to other object detection tasks, and turns the text area detection task into an object sensing task. Thus, the proposed text to be observed directly in the neural network’s convolutional characteristic map, and it can simultaneously predict the text/non-text score of the proposal and the coordinates of each proposal in the image. Then, we proposed an algorithm for the construction of the text line, to increase the text detection model accuracy and consistency. We found that our text detection operates accurately, even in multiple language detection functions. We also discovered that it meets the 2012 and 2014 International Conference on Document Analysis and Recognition thresholds of 0.86 F-measure and 0.78 F-measure, which clearly shows the consistency of our model. Our approach has been programmed and implemented using Python programming language 3.8.3 for Windows.

Hybrid Technique: Text Detection Using a Neural Network and Boxes

المجلة: International Journal of Computing and Digital Systems

سنة النشر: 2021

تاريخ النشر: 2021-01-01

Text Recognition is one of the challenging tasks of computer vision with considerable practical interest. This paper presents a text detector system, named NNBoxes, which detects text with both high accuracy and efficiency and move with the topic forward, NNBoxes is a hybrid technique between the Neural network (Noise reduction and letters prediction) and (Lockbox and hitbox) method created by the Author, both boxes draw grids and lines to recreate the shape of the letters and draw a path between the boxes to support the decision algorithm, in some cases the boxes technique detect text pattern in the image that the language text is not supported in the algorithm Dataset and train data.

Recognition of Handwritten English Numerals Based on Combining Structural and Statistical Features

المجلة: IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING

سنة النشر: 2021

تاريخ النشر: 2021-02-04

Generally, pattern recognition considered a strong challenge in many information processing research fields. The aim of this paper is to propose a highly accurate model for recognizing a handwritten English numeral through efficiently extracting the most valuable features of a certain handwritten numeral or digit. The handwritten English Numerals Recognition Model (HENRM) is proposed in this paper. The features extraction of the proposal based on combining both statistical and structural features of the certain numeral sample image. Mainly, the proposed HENCM has four phases which are image acquisition, image preprocessing, features extraction, and classification. In fact, four feature extraction approaches are utilized in this paper, which are the number of intersection points, the number of open-end points, calculation of density feature, and determining the chain code for each of the English numerals. The latter phase gives a features vector of 26-element size to be fed into the classifier that uses the Multi-class Support Vector Machine (MSVM) for the classification process. The experimental results showed that the proposed HENCM exhibits an average recognition rate equals to 97%.

Travellers decision making through preferences learning: A case on Malaysian spa hotels in TripAdvisor

المجلة: Computers & Industrial Engineering

سنة النشر: 2021

تاريخ النشر: 2021-04-20

Tourism has been one of the biggest competitive industries in the world. Nowadays, medical and wellness tourism are quickly developing as a part of tourism for health and wellness care. Social networking sites have played an important role in developing these types of tourism. Online reviews on the tourism products in social networking sites are considered rich sources for tourists’ decision making. Machine learning techniques have proved to be effective in analysing the tourists’ online reviews. For big datasets of tourist online reviews, these techniques must be enough robust to accurately discover the hidden relationships of tourists’ preferences in the online reviews. In addition, scalable machine learning techniques are needed for examining big datasets analysis in tourism platforms to timely provide the required information regarding the tourists’ preferences on the products. This paper investigates the effectiveness of a hybrid method using clustering, Higher-Order Singular Value Decomposition (HOSVD) and Classification and Regression Trees (CART) in analysing tourists’ online reviews in TripAdvisor. We use HOSVD to find the similarities among the travellers in the datasets with huge sets of hotels ratings. Then, we use CART to predict travellers’ preferences on the quality dimensions of spa hotels in TripAdvisor. To evaluate the method, the data is collected from the travellers’ online reviews on Malaysian spa hotels in TripAdvisor. The results showed that our method outperforms the methods which solely rely on prediction machine learning techniques. We demonstrate that the use of clustering and prediction machine learning techniques combined with the HOSVD is robust in analysing the tourists’ online reviews for discovering the tourists’ preferences in social networking sites.

Adaptive Filter based on Absolute Average Error Adaptive Algorithm for Modeling System

المجلة: UHD Journal of Science and Technology

سنة النشر: 2022

تاريخ النشر: 2022-05-07

Adaptive identification of the bandpass finite impulse response (FIR) filtering system is proposed through this paper using variable step-size least mean square (VSS-LMS) algorithm called absolute average error-based adjusted step-size LMS as an adapted algorithm. This algorithm used to design an adaptive FIR filter by calculating the absolute averaged value for the recently assessed error with the previous one. Then, the step size has been attuned accordingly with consideration of the slick transition of the step size from bigger to smaller to score an achievement through high convergence rate and low steady-state misadjustment over the other algorithms used for the same purpose. The simulation results through the computer demonstrate remarkable performance compared to the traditional algorithm of LMS and another VSS-LMS algorithm (normalized LMS) which used in this paper for the designed filter. The powerful of the filter has been served in the identification system, bandpass filter has been chosen to be identified in the proposed adaptive system identification. It reports conceivable enhancements in the modeling system concerning the time of convergence, which is well-defined as a fast and steady-state adjustment defined with a low level. The designed filter identified the indefinite system with less than 10 samples; meanwhile, other algorithms were taking more than 20 samples for identification. Alongside the fine behavior of preserving the tradeoff between miss adjustment and the capability of tracking, the fewer calculations and computations regarding the algorithm requirement make the applied real-time striking.

New Feature-level Algorithm for a Face-fingerprint Integral Multi-biometrics Identification System

المجلة: UHD Journal of Science and Technology

سنة النشر: 2022

تاريخ النشر: 2022-02-11

This article delves into the power of multi-biometric fusion for individual identification. a new feature-level algorithm is proposed that is the Dis-Eigen algorithm. Here, a feature-fusion framework is proposed for attaining better accuracy when identifying individuals for multiple biometrics. The framework, therefore, underpins the new multi-biometric system as it guides multi-biometric fusion applications at the feature phase for identifying individuals. In this regard, the Face-fingerprints of 20 individuals represented by 160 images were used in this framework . Experimental resultants of the proposed approach show 93.70 % identification rate with feature-level fusion multi-biometric individual identification.

Offline Handwritten English Alphabet Recognition (OHEAR)

المجلة: UHD Journal of Science and Technology

سنة النشر: 2022

تاريخ النشر: 2022-08-20

In most pattern recognition models, the accuracy of the recognition plays a major role in the efficiency of those models. The feature extraction phase aims to sum up most of the details and findings contained in those patterns to be informational and non-redundant in a way that is sufficient to fen to the used classifier of that model and facilitate the subsequent learning process. This work proposes a highly accurate offline handwritten English alphabet (OHEAR) model for recognizing through efficiently extracting the most informative features from constructed self-collected dataset through three main phases: Pre-processing, features extraction, and classification. The features extraction is the core phase of OHEAR based on combining both statistical and structural features of the certain alphabet sample image. In fact, four feature extraction portions, this work has utilized, are tracking adjoin pixels, chain of redundancy, scaled-occupancy-rate chain, and density feature. The feature set of 27 elements is constructed to be provided to the multi-class support vector machine (MSVM) for the process of classification. The OHEAR resultant revealed an accuracy recognition of 98.4%.

Construction of Alphabetic Character Recognition Systems: A Review

المجلة: UHD Journal of Science and Technology

سنة النشر: 2023

تاريخ النشر: 2023-02-18

Character recognition (CR) systems were attracted by a massive number of authors’ interest in this field, and lot of research has been proposed, developed, and published in this regard with different algorithms and techniques due to the great interest and demand of raising the accuracy of the recognition rate and the reliability of the presented system. This work is proposed to provide a guideline for CR system construction to afford a clear view to the authors on building their systems. All the required phases and steps have been listed and clarified within sections and subsections along with detailed graphs and tables beside the possibilities of techniques and algorithms that might be used, developed, or merged to create a high-performance recognition system. This guideline also could be useful for readers interested in this field by helping them extract the information from such papers easily and efficiently to reach the main structure along with the differences between the systems. In addition, this work recommends to researchers in this field to comprehend a specified categorical table in their work to provide readers with the main structure of their work that shows the proposed system’s structural layout and enables them to easily find the information and interests.

Kurdish standard EMNIST-like character dataset

المجلة: Data in Brief

سنة النشر: 2024

تاريخ النشر: 2024-01-09

A dataset was created by collecting handwritten samples of distinct Kurdish characters. The dataset consists primar- ily of 58 characters, and approximately 3800 adult volun- teers who are native Kurdish speakers participated in the col- lection process. Each participant was requested to fill two rows in a character form printed on A4 landscape papers. These papers were divided into sets of four pages, with 18 columns and 10 rows of characters on each page, except for the fourth page in each set, which had 40 cells. To ensure a comprehensive dataset, over 760 sets were prepared and distributed across various universities and institutions. The collected samples underwent scanning, cropping, and prepro- cessing procedures following the characteristics established by the EMNIST project. The purpose of these procedures was to standardize the dataset and ensure uniformity in the rep- resentation of all characters.

OFFLINE KURDISH CHARACTER HANDWRITTEN RECOGNITION (OKCHR) USING CNN WITH VARIOUS PREPROCESSING TECHNIQUES

المجلة: Journal of Engineering Science and Technology

سنة النشر: 2023

تاريخ النشر: 2023-12-14

Handwriting character recognition is an active and challenging area of research in the pattern-recognition field. Nowadays, many algorithms are introduced to achieve the highest accuracy. An improved recognition result could be accomplished if the input character images have good quality. That is why the pre-processing step becomes crucial for image identification missions. The algorithm of CNN (Convolutional Neural Networks) is fortified with numerous designs, permitting researchers to select the most operative architecture for better classification. However, this study suggests that the pre-processing mechanism is an important factor to be considered to increase the identification accuracy. This study utilised seven stages for pre-processing mechanisms applied to the dataset then the classification-accuracy measurement resultant from CNN was acquired. This work exposed that the utilisation of pre-processing steps indicated the most heightened accuracy with Binarized-KDS as an input, achieving training, testing, and validation accuracy of 99.2%, 97%, and 97.2% respectively after 35 iterations. Furthermore, another significant finding is that using different steps in processing input images to train the recognition model affects the recognition- rate within the same classifiers. Besides, the outcome reveals that each technique applied with the specific classifier may require a certain pre-process to obtain its optimal accuracy recognition rate.