Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

i need help on image processing

Status
Not open for further replies.

joanna

New Member
gudpm. i am currently a fourth year ECE student in XavierUniversity. our proj proposal is due this week and i am planning to do image - processing...machine vision... i just don't know where to start. i am really interested in this area and i would like to know more about it. i really hope you, guys, can help... tnx a lot! :)
 

bmcculla

New Member
Machine vision is a very broad topic. The most common use for machine vision is in quality control - make sure your product looks the way is should before it's sent out. You could use any number of defect detection applications for as project(bent IC pins identification, pill/m&m candy color Identification, coin sorting). Another interesting problem is character recognition, when a computer identifies characters in an image and converts them to ascii text.

The basic algorithms involved in machine vision are fairly simple. Most involve a 3x3 matrix of pixel scaling values. The matrix works like a 2 dimensional FIR filter.

Getting good tools will be tough for machine vision applications. Tools tend to be very expensive. If you have access to Labview or matlab both have good image processing capabilities. You could write everyting in C++ or Visual Basic but it'll be a fair amount of coding. Cameras and image capture hardware is quite expensive. If you could buy a cheap digital camera and take test pictures to run through your algorithims. The Character identification would be the easiest problem as far as tools are concerned because you don't need a camera you could use screen captures on your computer.

Hope this gives you a start on where to look.
Brent
 

joanna

New Member
tnx

tnx for taking the time to reply. one more question, can you suggest any site that could be very helpful for young hopefuls like me? i tried surfing the net but it seems like the site is for experts and i am actually at a loss to what they are actually saying. pls. help... tnx very much.
 

motion

New Member
tnx for taking the time to reply. one more question, can you suggest any site that could be very helpful for young hopefuls like me? i tried surfing the net but it seems like the site is for experts and i am actually at a loss to what they are actually saying. pls. help... tnx very much.


Here is a site that demonstrates what image processing can do. They have demo software that can read the 2D DataMatrix barcodes found on many products like the Pentium 4 processor. You can take almost any image downloaded from the web and can decode it. Sample images can also be found from this site.

Hopefully it will lead you to a specific project where you can apply image processing.

https://www.intelcom.ru/2d/english/demo.php

https://www.intelcom.ru/2d/english/applicat.php?group_id=a3
 

know_none

New Member
www.mathworks.com

>>>Matlab site! Click SUPPORT link to browse documentation.

Matlab is great! we did a project on photomosaics using this software.. we used very basic functions too. i see lots of thesis groups using Matlab for their DSP needs. matlab has an impressive image processing toolbox that you can check out... i think it will really help if you visit the site and also look for 'm-files' on image processing-these can really help you on your source code. good luck on your project.
 

joanna

New Member
tnx everyone! i think i want my robot to do just a simpe color object detection. i am still in the dark about this topic and i really appreciate all the help i could get. tnx very much! :D
 

bmcculla

New Member
There are a million way to identify colored objects. Heres one to get you started:

Most color images are represented as a 24 bit RGB (red green blue) array. Each color has 8 bits representing its intensity. if you are doing color detection and you want to identify a green object you just throw away the other two colors and you are left with an 8 bit green image.

A basic image manipulation algorithm, called threshold, is a simple comparison. if a pixel is higher than a specific value you output a 1 and if it is less than that value you output a 0. This give you a binary image in which the greenest places are 1's. A binary image can be viewed as a black and white image in which all the pixels are either black (1) or white (0) with no greys. Binary images are really easy to process.

now you want to eliminate all the areas that are too small to be the object you are looking for. An algorithm known as "dissolve" which as you might guess removes the outermost layer of 1's from each region. You keep "dissolving" the image until all the areas too small to be your object have dissapeared - whatever is left is likely to be your object. Dissolve works by looking at the pixels around each pixel and the pixel itself. If any of these 9 pixels arn't 1 that means the center pixel is on the edge of an object and should be set to 0 in the output.

Hope this helps.

Brent
 
Status
Not open for further replies.

Latest threads

EE World Online Articles

Loading
Top