Do any of you have experience with cheap optics? I have an inexpensive serial JPG camera that is sensitive to IR light, also a 2w IR laser. During the day I can easily sense with code on the Arduino when there is significant motion/change in the image. I can do this at least 5fps. Great but of course it doesn't work in the dark. With the laser and a servo, I can see when the laser hits a nearby object in the dark, compared to when it hits one far away like 200m. In the latter case there is no red dot. The further away it is, the larger the dot from 1-20m. When it hits an object close, it only changes 1 pixel on my 320x240 image. It is hard to detect this change without uncompressing the entire JPG image. This is much too slow. The camera has a cheap plastic lens.
What kind of optics can I use to change the single sharp laser dot to a large cross or X?
What is the word for this optic effect?
I can do it now when I squint at a bright light in the dark. If I can achieve this with cheap optics it will be much easier to detect a laser target than looking for 1 bright pixel.
Is there a different sensor, other than a camera, which can detect this binary result in the dark?
I just need to detect if there is a difference when the laser is on vs off, the answer I need is yes or no at 5fps.
It's either a red dot or complete darkness. I do not need to know the x,y coordinates or angle since I already know the angle of the servo with the laser on it. At this time I don't even need to know the size off the dot, unless I want to estimate the distance. I do not need to measure the intensity, it's all or nothing for me. Any ideas? Outside of the box?
What kind of optics can I use to change the single sharp laser dot to a large cross or X?
It is called a diffraction grating and is essentially a pattern of lines.
A diffraction grating takes the two dimensional Fourier Transform of the spot of light to produce the shape. Many shapes can be produced with the correct grating. http://cns-alumni.bu.edu/~slehar/fourier/fourier.html
You might try something infront of the lens that will slightly distort the incomming light and enlarge the dot. Another lens that would cause the cam to be a little out of focus, something generally clear but would make the picture a little fuzzy (vaseline on cam lens use to be a method to "soften" images), or even vibrating the cam.
2 great ideas! Where can I buy a cheap diffraction grating? Which one of these will work for my application? IR range?
I think blurring the image will just make it harder to detect because then the edges won't be as sharp. Instead I like the idea of vibrating the camera. I could stop the laser from moving for 200ms and pulse it 4x 50ms on/off to differentiate it from any other dot of light already existing in the image. Or just leave it on while rotating the camera 60 degrees. Now I will have a horizontal line all the way across the image. Combining this with a diffraction grating the change will be obvious! Any improvements to this plan?
What would happen if I blink the laser at 2ms while turning the camera 60 degrees in 200ms? Approximately every other pixel will be dark then light! If the exposure time was short I may only capture a few cycles. That is much better than 1 pixel that I started with.
You guys are great! Very helpful. How does the Karma thing work on here?
@zoomkat- I don't want to vibrate the laser much because I want to measure the angle or location of the target. Sure I could put a tip on my 2w laser, but that would just make it weaker with less distance. It blocks part of the beam.
@tumbleweed- I don't want to just vibrate randomly in a circular pattern, that would simply make the dot bigger and less powerful/distance. Instead I want to make a line which is easily done with a stepper motor since the camera is so lightweight.
Is your laser on top of the camera(or atleast behind the field of view) while the actual red dot is being projected out into the fov of the camera? Also, is just the laser moving back and forth across the fov, or are the camera and the laser both moving where the dot stays in the middle of the view?
if your looking for whether or not the dot was seen compared to one where it isn't, wouldn't an image from the dark and no dot just be 0, while a dot in the image would be greater than 0? Depends on how many other light sources it catches I guess. There should at least be a check-sum of sort that would change, instead of looking through the entire image, right? Just an Idea, I don't know how exactly how an image is put together in jpeg format.
And just as an expirement, I created two images, one just black, and one with a red dot. Interestingly the one with out the red dot is 6.98kb, while the one with the dot is 7.21kb. So maybe the size difference could work as well? Maybe just check each new frame with the last frame and see if the size has changed a certain amount since you last checked it.
Great ideas Neato! First to answer your questions. Since I can sample at 5fps I want to move the angle of the laser about 0.5-1 degree each sample. I want to detect a new object and the size of existing objects. During each 200ms I can move the camera independently as explained above. If I only use 1 servo then the laser would always stay about in the middle of the image. In this case I could uncompress the image and look at a few pixels for 1 that is brighter. But uncompressing is much too slow on Arduino. The problem is the noise in the camera. In the dark there are many bright pixels, but not as bright or wide as a laser. Checking the size is a great idea! But unfortunately it doesn't work because of the noise in a black image. The size doesn't change at all. We think alike, but I already tried it sorry. With the new ideas from others above, I should be able to just look at size!
