Macine Vision Plan

Working alone and most of the time at home, the team doesn't get to know what all I am doing and how far I have progressed,and since I have fallen prey to this blogging virus I think its time to post something regarding that.
This is a short report on my progress and the plan I will be following.

I have divided the entire task into three main phases :
Phase 1
Implementation of IP algorithms to segment the image into its component objects.

This was done with the help of the color data and then using their relationship with the other objects in the image. The techniques used for this process are not specific and differ for every problem parameters. After narrowing down to two algorithms to process the color data, 1 of them is working properly and giving satisfying results. The second step was completed with the help of a very popular greyscale segmentation technique.
The last step in this phase is to cut and scale the identified blobs from the image.

Phase 2
Using Neural Networks to identify the target alphanumerics and color

The neural networks will be trained with the input data in the form of a vector,from the last phase. And the output will be printed or sent to the VDU.
Once the nets are working fine in MATLAB, the code will be ported to C/C++ .

Phase 3
Integration of all the hardware : camera,wireless adapters,converters

This will be done as soon as all the products are available in the lab.

Rohit Arora aka. RA