I want to do a project in image processing or rather video processing.
The idea is that a camera would be mounted on a bot and it will compare the videos that it records to videos that have been previously stored in memory and accordingly move in the corret direction.
Can anybody help out and tell is it possible in practical? if yes then any help would be much appreciated.
How abt RF cameras? can they be used to transmit videos? cause wifi costs are higher and hence reduces the project efficiency and also its practical implementation in market. any other links can be used?
Breaking down the video into single frames will work. So ill be able to compare these images with the one's stored in the memory.
Also wanted to know if there are any methods for compression of these images so that the data transfer can be fast because these images have to be compared to the databse and then the Robot should be sent the information - in which direction to move.
If single frames will work, you might consider nRF24L01+ transceivers, for example:
**broken link removed**
They are fairly inexpensive and offer a fairly good bandwidth.
As far as compression there are several commonly used compression algorithms, you'll need to determine whether lossy or lossless compression is viable for your application. There are cameras available which transmit serially a compressed image in JPG format.
I've purchased the above model for as little as $35 in low quantities. Make sure you select a camera with a TTL serial interface, not an RS-232 or RS-485 which would require additional components to interface with the microcontroller.
Could anyone suggest some good articles/papers or journals (Bachelor level) in this or related topics and also how DCT (Discrete Cosine Transform ) can be used here for compression?
Also any help regarding the use of Matlab in this work would be greatly appreciated. Thanx :smile:
In my view....you first take the video of the same scene two times either by wire or wireless manner .....and try to see what your are getting is the co-relation between two videos eighter frame wise compare or as bigdog said object tracking etc....bcz if this philosophy works in your system constraints then only..... there is point to move forward on this bcz many times the scene propensities like brightness, sharpness of the object, zoom of the camera etc....badly affect the system performance....
Thanks bigdogguru and milind.a.kulkarni. Working on that
Also working on another project in parallel so that can finalise one of the two.
The 2nd project is to make a bot follow a path using image processing. Would appreciate your inputs on this idea. Thank you.