Line Scan Camera for less than £30 ?

In summary: You can use a normal camera that can see in two dimensions, or use a cheap line scan camera.A line scan camera will see a lot of things that it will regard as a finger. But you can learn how to use a two dimensional camera, what a finger looks like: ∩. So it must see a half circle with two parallel tangents added: That's a finger! This is not a finger: ∧.The resolution you want is up to you. Just google "cheap line scan ccd."You can use a normal camera that can see in two dimensions, or use a cheap line scan camera.
  • #1
Sorade
53
1
Hi all,

I'm looking to get started with a multitouch project using stereovision. Since I'm only interested in 2D for the moment I'm looking for a cheap line scan camera that I could use for the project. Ideally with an Arduino or that I can directly plug in the computer.

The idea is to put 2 cameras in the top corners of the diplay that will be able to detect a finger touching the screen. Using the length of the finger image and a bit of trigonometry I can work out the position of the finger.

More information on the project can be found here: https://www.physicsforums.com/threads/optical-touch-screen-code.825527/#post-5188484

Does anyone know where I could find such a camera. Also what is the difference between a line scan camera and a line scan sensor ?

Cheers !
 
Engineering news on Phys.org
  • #2
Why don't you use a "normal" camera that can see in two dimensions? A one dimensional camera will see a lot of things that it will regard as a finger. But you can learn a two dimensional camera, what a finger looks like: . So it must see a half circle with two parallel tangents added: That's a finger! This is not a finger:

Due to the background in the picture, the one dimensional camera may not see an edge of the finger, but the two dimensional camera will.

Say you want the camera to find a circle : O , but due to a missing edge it will see: C :
The one dimensional camera, scanning horizontally, will not find the circle, but the two dimensional camera easy finds it by means of a Hough transformation. ( Look it up ).
 
Last edited:
  • #3
What resolution do you want? Just google "cheap line scan ccd"

http://www.robotshop.com/en/parallax-linescan-imaging-sensor-daughterboard-tsl1401.html?gclid=CKSo5qqglMcCFQEcaQod6YkCcw
 
  • #5
Thank you for your replies.

Why don't you use a "normal" camera that can see in two dimensions?

My programming skills are very limited so I though that dealing with less inputs would be easier.

My idea was that I would be able to get a row of values that I could then interpret: say 0000148410000000000000000 , each number is the value at one pixel along the line camera. I can then work out that the finger is located at the 14841 and centred on the 8. I'm aware that's a very simplistic view and that I will have to filter out the background noise but I thought it could work. Then, knowing the focal distance of my camera, the angle of my camera axis (45degrees from my x/y display coordinates), and the position of the finger on the image I can work out where it is using two cameras. (see here for more details: https://www.physicsforums.com/threads/optical-touch-screen-code.825527/#post-5188484).

One thing I'm not sure about is will the camera be able to focus on fingers just in front of it (say 5 cm away) and at the other end of the display diagonal (say 85cm). or will I get blurry image at the extremes ? say 0011144488444111000000000 which would make the detection of more than one finger more inaccurate, as well as locating the actual centre of the touch.
 
  • #6
I'm worried about your span: 5cm . . 85cm: Yes, you will get some very blurred fingers, and say that the 5cm-finger takes up 14 pixels (11144488444111) then the 85cm-finger will take up 0.82 pixels ( or is it just noise? ).

How do you know, that your pixel-patterns will be symmetric? Remember that the image is a result of a shadow/reflected light. Your attached link suggests no light-setting:

No light → no image.

I could suggest parallel light, created by some small light-bulb and a freshnell-lens. As this parallel light meets frosted glass, the shadow from a finger will always be seen sharp, and the distance between the camera and the frosted glass is constant.

Sunshine is parallel due to the distance of the sun, thus a sharp shadow.
 
Last edited:
  • #7
I'm not sure what is needed to be able to focus over a span of 80cm. I'm worried that a mechanical lens will just not cope with the rate at which it has to adjust to gather data (30 FPS or around that).

For the light issue, I thought that ambient light would be enough, just as for us when we turn the lights one. Another option I thought of would be to have a layer of glass above the display and shine IR LED through it. The camera field of view will be just above the glass but it won't detect anything since the IR will be dispersing upwards mostly, yet when a finger is put close to the glass its tip will be illuminated by the scattering IR and be more visible to the camera. A bit like looking at a screen from the side, you don't see the light coming out of it that much, but when an object is put in front you can see the light on it. Alternatively the light from the display itself might be enough.

I wonder how they got around the focus problem in the paper though ?
 
  • #8
Sorade said:
Alternatively the light from the display itself might be enough.
But emits your display/screen constant light?

It's easy to test the setup: Just act a (stupid) camera at different circumstances ( different light/background, black & white color, and so on ):

Can you always sort out a finger? Forget during the test your know-about that it is your finger you are looking at. :smile:
 
  • #9
Hesch said:
Forget during the test your know-about that it is your finger you are looking at.
And remember to look through a crack.
 
  • #10
Hesch said:
freshnell-lens.
Sorry: fresnel-lens.

upload_2015-8-6_17-32-38.jpeg
 

1. What is a line scan camera?

A line scan camera is a type of digital camera that captures images one line at a time, rather than all at once. This allows for high-speed and high-resolution imaging of moving objects.

2. How does a line scan camera work?

A line scan camera uses a linear array of photosensitive sensors to capture an image one line at a time. The image is then reconstructed by combining all the lines captured in sequence.

3. What are the advantages of using a line scan camera?

Line scan cameras are ideal for capturing images of moving objects or large objects that cannot fit in a traditional camera's field of view. They also offer high-resolution imaging and can be used for quality control and inspection purposes in industries such as manufacturing and healthcare.

4. How much does a line scan camera cost?

A line scan camera can range in price from a few hundred dollars to thousands of dollars, depending on the resolution, speed, and other features. However, you can find some line scan cameras for less than £30, usually with lower resolutions and simpler features.

5. Can a line scan camera be used for scientific research?

Yes, line scan cameras can be used for scientific research in various fields, including biology, physics, and engineering. They can capture high-quality images of moving objects, making them useful for studying fast-paced phenomena. However, for more advanced research, a higher-end line scan camera may be necessary.

Similar threads

  • Electrical Engineering
Replies
8
Views
841
Replies
10
Views
2K
  • Electrical Engineering
Replies
7
Views
2K
  • Computing and Technology
Replies
10
Views
2K
Replies
2
Views
1K
Replies
1
Views
905
Replies
7
Views
1K
Replies
5
Views
4K
  • Electrical Engineering
Replies
18
Views
11K
  • Electrical Engineering
Replies
2
Views
2K
Back
Top