Hi everyone! I’m back with another webcam project, as usual. :) I’ve been having a lot of fun making cool things using my webcam over the last couple of months. This time, I’ve created an obstacle detector which uses a cheap laser pointer and a webcam to determine how far an object is from the camera. I worked with my cousins, Anuj Karpatne and Prateek Raj on this project.
There are all sorts of range finding devices available in the market like ultrasonic and infrared range finders. All these devices work well, but where portability and cost are of primary concern, this technique could be more desirable.
So, here’s how it works. A laser pointer is fixed below the camera at a known distance (as you can see in the picture above). The pointer is kept parallel to the optical axis of the camera. Then, a laser-beam is projected on an obstacle which is supposedly in the camera’s field of vision. The scene, along with laser dot is captured by the camera. Then, the image is scanned for the brightest pixels (we are assuming that the laser dot is the brightest point in the scene). Then, we calculate how far along the y-axis the laser dot is from the center of the image. The farther from the center, the closer we are to the obstacle. So, the distance from the object (D) is inversely proportional to the number of pixels from the center of the image (y):
D = hY/2y * cot (A/2)
Where the constants used are:
h = the actual distance between the laser pointer and the optical axis of the camera.
Y = total number of pixels along the y-axis
A = angle of view of the camera
We used an experimental approach to determine the value of k. We wrote a simple program in C# to do the entire image processing work and for finding the position of the laser dot in the scene. We calibrated the apparatus by placing an obstacle at known distances from the center of the camera and noting down the number of pixels the laser dot is from the center of the image along the y-axis. After taking several readings, we obtained the value of k. By using this value, we were actually able to determine how far an obstacle is, from the camera.
Well, we had a lot of fun working on this project! I hope you find it interesting. I’d love to hear your comments and ideas. :)
So, here’s how it works. A laser pointer is fixed below the camera at a known distance (as you can see in the picture above). The pointer is kept parallel to the optical axis of the camera. Then, a laser-beam is projected on an obstacle which is supposedly in the camera’s field of vision. The scene, along with laser dot is captured by the camera. Then, the image is scanned for the brightest pixels (we are assuming that the laser dot is the brightest point in the scene). Then, we calculate how far along the y-axis the laser dot is from the center of the image. The farther from the center, the closer we are to the obstacle. So, the distance from the object (D) is inversely proportional to the number of pixels from the center of the image (y):
D = hY/2y * cot (A/2)
Where the constants used are:
h = the actual distance between the laser pointer and the optical axis of the camera.
Y = total number of pixels along the y-axis
A = angle of view of the camera
As these values are constants, we obtain,
D α 1/y
Or, D = k/y (where k is the proportionality constant for all the constants used above)
D α 1/y
Or, D = k/y (where k is the proportionality constant for all the constants used above)
We used an experimental approach to determine the value of k. We wrote a simple program in C# to do the entire image processing work and for finding the position of the laser dot in the scene. We calibrated the apparatus by placing an obstacle at known distances from the center of the camera and noting down the number of pixels the laser dot is from the center of the image along the y-axis. After taking several readings, we obtained the value of k. By using this value, we were actually able to determine how far an obstacle is, from the camera.
Well, we had a lot of fun working on this project! I hope you find it interesting. I’d love to hear your comments and ideas. :)
143 comments:
Post a Comment