News:

MASM32 SDK Description, downloads and other helpful links
MASM32.com New Forum Link
masmforum WebSite

dsp and control processing

Started by ozzy_85, May 17, 2006, 09:10:56 AM

Previous topic - Next topic

ozzy_85

guyz can anybody give me the exact difference between digital signal processing and control processing, i'm getting confused here... :(

Ossa

Digital Signal Processing (DSP) deals with, well, signals. There are many many examples of DSP areas, here are a few examples:


  • Image processing - take noisy/blurred or otherwise imperfect images and reconstruct the original or desired image
  • Image coding/compression - take a digital image and make it smaller (in terms of bits) e.g. JPEG
  • Audio processing - reforming "perfect" audio from noisy/bad data
  • Audio compression - makeing audio smaller e.g. MP3

Here are some more (a bit more specific here):


  • Processing radio signals (e.g. in mobile phones), reconstruct the underlying signal from noisy measurements (using, many methods - simplest is Wiener filter), radio signals might be from actual radio, TV, mobile phones, satellite transmissions, radar - anything
  • Speech processing and coding - (A) compression/decompression e.g. LPC10, CELP; (B) text-to-speech, i.e. speech synthesis e.g. TDPSOA; (C) speech-to-text, the desktop dictation type things - very flexible, many combinations of solutions (for example, maybe use Mel-Frquency Cepstral Coefficient (MFCC) front end, a triphone model including cross word triphones and backoff, with a trigram language model and based on the Viterbi algorithm using beam search and token passing)
  • Computer vision - again, a million different applictions, structure from motion (make 3D models from a video), stereoscopic vision (2 or more cameras), mosaicing ('stitching' images together),...

the list goes on and on - basically DSP is the processing of information in the form of a signal (the signal being whatever you like - radio, audio, video). It often requires a great deal of maths to understand why things work (it's mostly frequency domain things - using Fourier transforms, DFTs, DCTs, Laplace, Z-tansforms).

Here's what wikipedia has to say on it: http://en.wikipedia.org/wiki/Digital_signal_processing

Now... I'm not too sure what you mean by "control processing", but here are my two guesses and I'll explain both.

1) Microcontroller style control of a system. The processor simply responds to stimuli to produce outputs. Examples are in almost anything electronic that you use. Often uses rules defined by (2) to issue it's outputs.

Wikipedia says: http://en.wikipedia.org/wiki/Microcontroller

2) Control engineering type control. This is concerned with the design of mathematical rules that allow a system to do as desired. As a simple example, the thermostat in a house controls whether the heater is on or off the rules are that if it is too hot, turn it off, if it is too cold, turn it on. As a much more complex example, the system that converts a pilot's "requests" to a jet fighter into movements of the control surfaces is a non-linear multivariable control problem and uses many much more complex mathematical rules. To understand this type of control, you basically need to take a degree in it. It involves a great deal of mathematics (take just one non-linear stability criterion - the circle criterion - it needs understanding of state-space descriptions, which in turn needs calculus. Or the newer style robust linear controller types - also need (at the very least) LaPlace and Z-transforms to be understood well).

Here's what wikipedia has to say: http://en.wikipedia.org/wiki/Control_theory
and: http://en.wikipedia.org/wiki/Control_system

Not sure if that answered your question or not, but hope it helps,
Ossa
Website (very old): ossa.the-wot.co.uk

ozzy_85