​ What's the best way to process image and video right from a CMOS sensor? COMPUTER SCIENTISTS, I NEED HELP!

[Caio Macedo]

I love to understand computation deeply, right from the CPU architecture, and I am having some problems with an embedded camera. I don't know what's the best way to do it, I know that microcontrollers aren't the best option, because they can't deal with all that data (4k/ 8k / 1080p 60fps - high resolution images) I know that the first circuitry that connects to the silicon sensor chip are specific driver chips to provide clock and send the data from all the matrices of pixels via parallel, so I've been reading a lot about Digital Signal Processor (DSP) and Digital Signal Controller (DSC), but I don't know how to connect the sensor, don't know if I only interface it with the DSP and send the data via USB/HDMI and process it in my computer (just another motherboard) with LibUsb, openCV and some graphics library in c; or interface the sensor, DSP and external ROM memory; even thought about interfacing external CPU (intel) proper RAM and ROM and then, interface the DSP and sensor, so that I can create a autonomous robot that process live video from the camera with algorithms created by me, and I don't worry about writing programs in assembly (if this is your solution).