Programming cameras to detect patterns

[Lucas.Steele]

Hello everyone, I am trying to make something that will help my brother. Long story short, he has a lazy eye. He can correct it when he realizes the eyes deviates, but obviously he cannot think about it every second of the day.
What I want to do is put two small cameras in a pair of glasses, so they can detect when one eye deviates from the other. I would like the trigger of detection something like a small red light turn on (that he can see, of course) when there is deviation.

I imagine this will is a fairly simply project, with the only real obstacle being the programming of the cameras to recognize the pupil from the rest of the eye and eyelids, and establishing the pattern between the two pupils.

I need to know the hardware necessary (Type of chip for the programming, type of cameras, battery, etc.),
The software to program the recognition of eye deviation (and how to program it onto a chip),
And how to connect all of these components.

Clearly, I don’t know much about electronics, but I will be researching all of this so I am not just blindly following instructions. I just need help with the basic information to point me in the right direction.

Please feel free to send me an email at Lucassteele101@hotmail.com
Thank you all in advance!

 

[barry]

I admire your attempt to help your brother out! However, I don't think this project is as simple as you think. First, you need to interface two cameras to some intelligence-I'm not an image processing expert, but I think you are going to need more than a simple microprocessor-probably and FPGA or DSP. This is the easy part. Then you need to develop some kind of algorthm that can do the following:

1) Identify the pupil in each eye. Deal with issues like glare, different lighting conditions, shadows, blinking, crying at sad movies, etc.
2) Determine where each of those pupils is 'aimed'
3) Determine if the axis of one pupil is skewed from the other axis.

This sounds like a pretty formidable image-processing system, even for an expert.

Perhaps others on this board can offer more hope than me! Good luck.

 

[BradtheRad]

An experiment that might be worth trying is to mount the cameras so that they produce identical images when the eyes are aimed parallel.

Convert the pixels of one image to negative, mathematically, by inverting their value.
Then add each pixel to the corresponding pixel in the other image.

When the eyes are parallel, the pupils coincide. The images are identical. In combining one image with the negative of the other, the result is zero.

When the eyes become skewed, one image will show some white where the other shows pupil. This creates a mathematical difference between them.

This method assumes that both eyes have equal amounts of light shining on them, etc.

You will need cameras that focus at a fraction of an inch distance.
You want images to consist only of eyes, with a minimum of eyelids.

 

[manvindar]

u can try reading this https://www.ti.com/solution/automot...-cc-3q13&HQS=analog-mcs-cc-4q13-fb-adascamera

it may help you to start