Best way for recognizing an object(help for making a device)

Hi everybody
i want to make a device untill recognize an object. for example: cube, sphere or etc
now, i have two sensor that they making pulse for measuring distance between sensor and object.
like this picture:



due to picture, the T4 is The width of the pulse and the T2 will change due to object.
i can recognize the object by this amounts. for this job, it just is necessary that first i put the objects one by one under the sensors and save the amounts of sensors(with the name of objects). next,
for exmaple, for recognizing an objects, i put the object under the sensors and the MCU(or processor) say me the name of that object.
the sensors is two number. the objects are 100 number. for each object i want to sampling 20 times. thus, for each object, i should save 40 amount.
which processor or microcontroller is suitable for this job? AVR, ARM based microcontroller, FPGA,...
any suggestion for do this job?

rozeh,


Take a look on this approach :
https://www.edaboard.com/threads/292972/


+++

Thanks andre,
but i don't think it can help me.

rozeh,


If you consider to use cheap platforms systems based on ARM11 or CORTEX-A8, it is possible to install embedded Linux.
Under this operational system, you can install some QT C++ compiler, and can run OpenCV examples.

There are at some examples based on this library based on image enancement and shape recognition.


+++