Vote for Matlab, because: are fully supported, there are many libraries, easier to handle more than others. If you want to convert to others, for example, C++ project, Matlab also could be useful.

Well, I recommend nodejs as a suitable technology for real time applications. Nodejs uses an event-driven, non-blocking I/O model that makes it lightweight and efficient, ideal to manage devices through sockets. Also, there are a lot of libraries like cylon.js which are compatible with a set of iot devices.

Moreover, there are some micro-controllers that support javascript as programing language. This feature, make possible to program all in one language.

What do you mean exactly as real-time image processing on IoT devices? These devices are generally much slower than computer CPUs and you will have a hard time getting anything but basic image processing anywhere close to 30fps.

For example, we developed a visual tracking algorithm running at more than 200fps on a recent Intel I7 system but it runs only at 3fps on the Hololens (which uses an Atom 1.1Ghz cpu) and is useless for tracking purpose in this device...

My suggestion:

- Decide on exactly which processing algorithms you will be using by prototyping them on a good PC keeping them as lightweight as possible. Without knowing exactly what algorithms you will run, it is very difficult to make a sensible choice of device.

- Remember that another strategy could be to transmit the image or video to a powerful workstation which could do the hard work much faster!

Finally, you might be better looking at some existing off the shelve commercial devices. For example:

https://q-see.com/collections/ip-hd-nvrs

C programming is suitable. I think it is not that much efficient to do image processing in Arduino. If you know Raspberry pi well, then you will go for that. Once we tried Open CV in Python, and that was easy.

Actually, if you know any language perfectly at any hardware platform, you will do very well.

Please follow the previous persons (Mr. Bruno & Abraham) answers

Thank you.

Vote for Matlab, because: are fully supported, there are many libraries, easier to handle more than others. If you want to convert to others, for example, C++ project, Matlab also could be useful.

It strikes me that what we need to know from Rashmiranjan Nayak what processing operates where? How do the end point on-camera units communicate with the surveillance system, in what form, and then what happens at the surveillance system or whatever the hub arrangement is.

The topology and the nature of the hops to where someone is watching digital screens and any added analysis seems important before choosing processors and languages.

I also recommend Bruno Martin's appraisal.

I'm vote to matlab as a programming language also you can generate C++ code from matlab when you used rasbeery pi 3 .

all the best