Hello, this is a good project! It is more machine / deep learning than image processing though. I did a similar project for one of my classes but it was only four months long. I basically had a webcam detect, track, and then identify objects moving across the cameras field of view. You can find code and more information here.
Just thinking of your project quickly I would look at AlexNet and do transfer learning on a dataset of cars. From there I would associate those images with some sort of data bank that has all the information like color, make, etc...
If you find it to easy or finish after four months there is lots you can add to the project to make it more complex. For example making a better classifier or your own CNN, optimizing computing time with the Parallel Toolbox (building your classifier is very computationally intensive), or even building a GUI to make it more user friendly.
If you have any other questions let me know but you can bring this information to your professor to see if its inline with your course objectives but I certainly think you can turn this into a great final year project.
