FPGA based visual recognition system for blind people
Blind people are known to make better use of their senses. Some of them can even determine the position of objects purely by using their voice (cfr. Daniel Kish, echolocation). But wouldn’t it be handy if they could use a system that converts information that they simply can not obtain into something observable? This is the goal of our thesis.
We are going to make an embedded system using an FPGA (=field programmable gate array) board that converts camera pictures into sound, so that blind people can sense visual information. The system we are using is a Digilent Nexys 3 board, based on the spartan 6 chip from Xilinx. This board contains hardware that can be programmed with VHDL (= VHSIC hardware description language) . Images are read in from an HDMI(=High-Definition Multimedia Interface) camera and stored in external DRAM(=Dynamic random-access memory memory). Afterwards some DSP(=Digital Signal Processing) calculations are done to convert the images in to sound. Finally PWM(= Pulse Width Modulation) is used to convert the digital signals into a sound source.
Edit: explained some terms & added foto of the Nexys board
Technology Blog 1315 
How does this embedded system look like? Has it such a size that people can put it into their pocket?
Can you please explain abbreviations like VHDL, DSP, …? And is it also possible to sense colors, or only shapes? In that sense, how about color-knowledge with blind people? 😀
This thesis is a thesis that a lot of student will be working on for the next years, so our task is to make a basic implementation that can be improved in the next few years. That implies that we are not going to make a finished product that must fit in a person’s pockets. This is however possible, because there are smaller boards that include the same spartan 6 FPGA, for instance the Avnet Spartan-6 LX9 MicroBoard (http://www.xilinx.com/products/boards-and-kits/AES-S6MB-LX9.htm). It is possible to detect shapes, light intensity and colors, but not all these features are going to implemented by us, for the reason that is stated above.
What would the generated sound sound like? Are there already some implementations on a maybe less mobile platform?
I’ve heard that there are a lot of systems that can convert text, maybe even pictures into Braille. Is there a specific advantage for the conversion into sound? Maybe these two systems can be combined?