2/13/2015 BIOMEDICAL RESEARCH: Image processing and neural networks assist the blind Vision Systems Design http://www.visionsystems.com/articles/2009/12/biomedicalresearchimageprocessingandneuralnetworksassisttheblind.html 1/8 BIOMEDICAL RESEARCH: Image processing and neural networks assist the blind Agerelated macular degeneration and retinitis pigmentosa are two of the major causes of loss of eyesight, leaving patients with difficulty reading, sewing, or discerning faces. In severe cases, these patients may lose their vision entirely. To aid those suffering from such conditions, Eng. Enas Elbarbary of VACSERA (Cairo, Egypt; www.vacsera.com), Ass. Prof. Hussam Elbehiery of the Air Defense Forces (Cairo, Egypt), and Prof. Abdelhalim Zekry and Prof. Adel Elhennawy of Ain shams University (Cairo, Egypt; www.shams.edu.eg) have developed a portable system, known as the Blind Assistant, to help the visually impaired recognize their surrounding objects and avoid dangerous situations. This automated system consists of a portable camera interfaced to a mobile computer that uses imageprocessing techniques and neural networks to detect objects and their colors and provide pathplanning aids. Using the MATLAB Image Acquisition Toolbox from The Mathworks (Natick, MA, USA; www.mathworks.com), images from the camera are first captured as objects where their settings can be edited and stored. To allow surrounding objects, characters, traffic signs, and currency to be interpreted, the system uses The Mathworks’ Neural Network Toolbox for pattern recognition, a separate algorithm to determine the color of an object, and template matching for path planning. By employing a sound card in the portable computer, specific Home NonIndustrial Vision BIOMEDICAL RESEARCH: Image processing and neural networks assist the blind HOME FACTORY AUTOMATION NONINDUSTRIAL VISION CAMERAS BOARDS & SOFTWARE PRODUCTS SUBSCRIBE | ADVERTISE
Transcript
2/13/2015 BIOMEDICAL RESEARCH: Image processing and neural networks assist the blind Vision Systems Design
BIOMEDICAL RESEARCH: Image processing and neuralnetworks assist the blind
Agerelated macular degeneration and retinitis pigmentosa are two of the major causes of lossof eyesight, leaving patients with difficulty reading, sewing, or discerning faces. In severecases, these patients may lose their vision entirely.
To aid those suffering from such conditions, Eng. Enas Elbarbary of VACSERA (Cairo, Egypt;www.vacsera.com), Ass. Prof. Hussam Elbehiery of the Air Defense Forces (Cairo, Egypt),and Prof. Abdelhalim Zekry and Prof. Adel Elhennawy of Ain shams University (Cairo, Egypt;www.shams.edu.eg) have developed a portable system, known as the Blind Assistant, to helpthe visually impaired recognize their surrounding objects and avoid dangerous situations. Thisautomated system consists of a portable camera interfaced to a mobile computer that usesimageprocessing techniques and neural networks to detect objects and their colors andprovide pathplanning aids.
Using the MATLAB Image Acquisition Toolbox from The Mathworks (Natick, MA, USA;www.mathworks.com), images from the camera are first captured as objects where theirsettings can be edited and stored. To allow surrounding objects, characters, traffic signs, andcurrency to be interpreted, the system uses The Mathworks’ Neural Network Toolbox forpattern recognition, a separate algorithm to determine the color of an object, and templatematching for path planning. By employing a sound card in the portable computer, specific
Home NonIndustrial Vision BIOMEDICAL RESEARCH: Image processing and neural networks assist the blind
instructions about an object, its color, and its distance are then relayed to the user.
To locate objects and determine the distance between them, the Blind Assistant system usestemplate matching to predict distances between objects and the number of steps required toreach them.
Click here to enlarge image
To locate objects, the Blind Assistant system uses template matching and predicts theapproximate number of steps required to reach them.
“For object recognition, neural networks must be adjusted or trained based on a comparison ofthe output image and the target image so that a particular input leads to a specific targetoutput. Because of this, developers deploying neural networks need not develop algorithms toperform specific tasks,” Elbarbary says.
2/13/2015 BIOMEDICAL RESEARCH: Image processing and neural networks assist the blind Vision Systems Design
Using a number of previously acquired images, Elbarbary trained the system’s neural networkwith objects consisting of characters, objects, currency, and traffic signs. To determine thecolor of an object, the average RGB values in the image are computed and compared withpreviously stored colors in a color library. While object recognition was achieved using apredictive feedforward neural network, template matching is used for objects locationdetermination. The approximate distance between the blind and the objects as well as thenumber of steps required to reach them are predicted (see figure).
“By using neural networks for object recognition, features within specific images can beautomatically trained and used to match future images as they are captured. Since trainingtime is not dependent on processing time, comparing a newly captured image to the trainedmodel, character recognition, and object recognition are relatively fast,” Elbarbary says. Byincreasing the number of these trained models, the likelihood of finding the captured objectusing the neural network is increased.
To determine the number of steps needed to reach an object, an approximate method is usedto predict them from captures. This is more computeintensive than using a neural network andincreasing the number of templates needed to determine different object orientations, scale,and accommodate changes in illumination or increases in the processing time required.
The researchers have tested the Blind Assistant system in both an office and exterior streetsettings using both neural networks and template matching for object recognition and for pathplanning. To image characters, objects, currency, and traffic signs, the average recognitiontime (per character) was approximately 0.5 s. Using the artificial neural network for objectrecognition took approximately 0.9 s. However, when template matching was used for pathplanning, the average processing time increased to approximately 1.7 s for long distances and7 s for short distances.
How to apply infrared imaging in vision systemsThis updated webcast begins with a review of the benefits and operational specifics of infrared imagingfor machine vision applications. Systems integration and engineering specialist Dr. Jonathan ...
UAV roundup 2/9: The latest in unmanned aerialvehicle news
UAV roundup 1/30: The latest in unmanned aerialvehicle news
Seeing through the Blizzard of 2015
Live updates from the 2015 A3 Business Forum
FAA grants additional exemptions for commercialUAVs
UAV lands on water, captures underwater footage
Student builds a camera inside a Lego
MORE
10 important considerations for designing a machine vision systemDeveloping a machine vision system is a complex activity involving optical engineering, electricalengineering, programming, and mechanical engineering. Moving from inception through installa...
Vision technologies for robotics: Application do’s and don’tsThis webcast will offer tips and examples for integration of machine vision systems in roboticsapplications. Expert Jeff Boeve of JR Automation will explain how to clearly define your pass/fa...December 9, 2014
Solving factory automation challenges with machine visionWhat do you need to know to implement your machine vision setup for industrial automation? Thiswebcast will answer that question using realworld application examples—such as inspection,assembly,...November 18, 2014
BLOG
TOPIC INDEX
View Vision Systems Design articles by topic, AZ
ARCHIVES
Click here to view archived Vision Systems Design articles
