Faruq A. Al-Omari

Handwritten Indian numerals recognition system using probabilistic neural networks

Abstract This paper presents a system for the recognition of the handwritten Indian numerals one to nine (1–9) using a probabilistic neural network (PNN) approach. The process involved extracting a feature vector to represent the handwritten digit based on the center of gravity and a set of vectors to the boundary points of the digit object. The feature vector is scale-, translation-, and rotation-invariant. The extracted feature vector is fed to a PNN, which in turn classifies it as one of the nine digits. A set of experiments were conducted to test the performance of the system under different angles between the vectors from the centroid to the boundary of the digit object. A 30° angle results in a 99.72% recognition rate with a short feature vector of 12 entries. This study is meant to be a seed toward building a recognition system for Arabic language characters.

Query by image and video content: a colored-based stochastic model approach

For efficient image retrieval, the image database should be processed to extract a representing feature vector for each member image in the database. A reliable and robust statistical image indexing technique based on a stochastic model of an image color content has been developed. Based on the developed stochastic model, a compact 12-dimensional feature vector was defined to tag images in the database system. The entries of the defined feature vector are the mean, variance, and skewness of the image color histogram distributions as well as correlation factors between color components of the RGB color space. It was shown using statistical analysis that the feature vector provides sufficient knowledge about the histogram distribution. The reliability and robustness of the proposed technique against common intensity artifacts and noise was validated through several experiments conducted for that purpose. The proposed technique outperforms traditional and other histogram based techniques in terms of feature vector size and properties, as well as performance.

IMPROVING GESTURE RECOGNITION IN THE ARABIC SIGN LANGUAGE USING TEXTURE ANALYSIS

Sign language plays a crucial role in communication between people when voices cannot reach them. Deaf people use sign language as their primary method of communication. Hand gestures represent the alphabets of sign languages. For proper inter-communication between hearing and deaf people, a translator becomes of great need. In this paper, a fully automated translator of the gestures representing the alphabets of the Arabic Sign Language (ASL) was developed. A set of 30 ANFIS networks were designed and trained properly to recognize the ASL gestures. The developed system is a visual-based system that does not rely on the use of gloves or visual markings. To this end, the developed system deals with images of bare hands, allowing the user to interact with the system in a natural way. A twin approach that is based on boundary and region properties is utilized to extract a set that recognizes the gesture. The extracted features are translation, scaling, and rotation invariant so as to make the system more flexible. The subtractive clustering algorithm and the least-squares estimator are used to identify the fuzzy inference system, and the training is achieved using the hybrid learning algorithm. Experiments revealed that our system was able to recognize the 30 Arabic manual alphabets with a recognition rate of 100% when approximately 19 rules are used per ANFIS model, and a recognition rate of 97.5% when approximately 10 rules are used.

A novel approach for quantitative assessment of mucosal damage in inflammatory bowel disease

AimsOne of the main reliable histological features to suggest the diagnosis of inflammatory bowel disease is the presence of significant distortion of the crypt architecture indicating the chronic nature of the disease resulting in mucosal damage. This feature has a considerable intra-observer and inter-observer variability leading to significant subjectivity in colonic biopsy assessment. In this paper, we present a novel automated system to assess mucosal damage and architectural distortion in inflammatory bowel disease (IBD).MethodsThe proposed system relies on advanced image understating and processing techniques to segment digitally acquired images of microscopic biopsies, then, to extract key features to quantify the crypts irregularities in shape and distribution. These features were used as inputs to an artificial intelligent classifier that, after a training phase, can carry out the assessment automatically.ResultsThe developed system was evaluated using 118 IBD biopsies. 116 out of 118 biopsies were correctly classified as compared to the consensus of three expert pathologists, achieving an overall precision of 98.31%.ConclusionsAn automated intelligent system to quantitatively assess inflammatory bowel disease was developed. The proposed system utilized advanced image understanding techniques together with an intelligent classifier to conduct the assessment. The developed system proved to be reliable, robust, and minimizes subjectivity and inter- and intra-observer variability.Virtual slidesThe virtual slide(s) for this article can be found here: http://www.diagnosticpathology.diagnomx.eu/vs/1797721309305023

Image-based discriminating morphological features for gastric atrophy assessment: a step to go further.

The aim of this study is to establish a basis for automated assessment of gastric atrophy according to the Updated Sydney System. We sought to minimize inter- and intra-observer variations in the application of the Sydney System. A total of 160 biopsies were examined by three pathologists and graded using the visual scale of the Updated Sydney System. A consensus was reached on 135 biopsies. Digital images were captured for the studied biopsies. Image processing techniques were used to extract four morphological features that uniquely discriminate each atrophy grade. The features are related to gland density and shape. To validate the reproducibility of these features, the K-Means clustering technique was used. We were able to grade the atrophy with an overall precision of 95.6%. Furthermore, the proposed features were able to distinguish four discrete grades without any significant overlap. This has not been achieved by previous studies.

An intelligent decision support system for quantitative assessment of gastric atrophy

Aims To build an automated decision support system to assist pathologists in grading gastric atrophy according to the updated Sydney system. Methods A database of 143 biopsies was used to train and examine the proposed system. A panel of three experienced pathologists reached a consensus regarding the grading of the studied biopsies using the visual scale of the updated Sydney system. Digital imaging techniques were utilised to extract a set of discriminating morphological features that describe each atrophy grade sufficiently and uniquely. A probabilistic neural networks structure was used to build a grading system. To evaluate the performance of the proposed system, 66% of the biopsies (94 biopsy images) were used for training purposes and 34% (49 biopsy images) were used for testing and validation purposes. Results During the training phase, a 98.9% precision was achieved, whereas during testing, a precision of 95.9% was achieved. The overall precision achieved was 97.9%. Conclusions A fully automated decision support system to grade gastric atrophy according to the updated Sydney system is proposed. The system utilises advanced image processing techniques and probabilistic neural networks in conducting the assessment. The proposed system eliminates inter- and intra-observer variations with high reproducibility.

Development of a CAD/CAM system for simulating closed forging process using finite‐element method

Purpose – The purpose of this paper is to study the effect of the height and diameter of the dies as well as work‐piece dimensions, on stresses and strains on dies in the forging process. This helps in developing a better understanding of the effect of process parameters. As a result, the manufacturing task could be accomplished with minimal number of trials.Design/methodology/approach – After determining the most influencing parameters on the forging process, the mechanical part is drawn, size of initial billet and shape of punch and die are also determined to build a finite‐element model to represent the process. Several outputs are taken as an indication for die wear and process performance. Finally, a computer numerical control (CNC) code to manufacture the selected die is generated.Findings – It was found that when the die diameter increases, the effective stress decreases. On other hand, it was found that the work required to finish the forging process is highly affected by the dimensions of work‐pi...

Fast Video Shot Boundary Detection Technique based on Stochastic Model

A video is composed of set of shots, where shot is defined as a sequence of consecutive frames captured by one camera without interruption. In video shot transition could be a prompt hard cut or gradual fade, dissolve, and wipe. Shot boundary detection is an essential component of video processing. These boundaries are utilized on many aspect of video processing such as video indexing, and video in demand. In this paper, the authors proposed a new shot boundary detection algorithm. The proposed algorithm detects all type of shot boundaries in a high accuracy. The algorithm is developed based on a global stochastic model for video stream. The proposed stochastic model utilizes the joined characteristic function and consequently the joined momentum to model the video stream. The proposed algorithm is implemented and tested against different types of categorized videos. The proposed algorithm detects cuts fades, dissolves, and wipes transitions. Experimental results show that the algorithm has high performance. The computed precision and recall rates validated its performance.

Hierarchal Spatio-color Image Indexing and Retrieval Based on a Stochastic Model

In this paper, a hierarchal image indexing and retrieval technique is proposed. Both color distribution and spatial information in an image are incorporated to index images in a repository. A set of global and localized features are extracted for the image based on a stochastic model. A hierarchal approach is then used to index images based on the derived features. The reliability and robustness of the proposed technique against common intensity artifacts and noise is validated through several experiments conducted for that purpose. The proposed technique outperforms traditional and other histogram based techniques in terms of feature vector properties, as well as retrieval performance.

Enhancing large image database indexing and retrieval performance through integrating database structure with image features

In this paper, a two-dimensional hash table structure based on image discriminator measure is proposed to design an image database indexing and retrieval system. Unlike many other techniques such as R-tree, FastMap, MetricMap and SparseMap where a false query result could happen, the results of a query of our proposed system can retrieve exactly same results as if we using full search for the query image in the image database. Henceforth, the designed system outperforms existing systems that are based on sequential search, OMNI-family and R-tree structures in terms of number of operations and precision.