1. Sampling and Digitisation

2. Digitisation Real-life images and sounds need to be digitised for computer representation. Turning an analogue or continuous signal into a digital signal. There are 3 stages to digitisation. –Sampling –Quantisation. –Coding.

3. Sampling 1011, 1100, 1101, 0000, 0010, 0100, 0101, 0101, 0101, 0101, 0100, 0010, 0000, 1101, 1100, 1011, 1011, 1011, 1011, 1100 Quantisation Coding The 3 stages of Digitisation

4. Sampling Examine the signal at discrete intervals (in space or time). Nyquist’s sampling criteria states that we must sample a signal twice as fast as it is changing to represent it accurately. If we do not we get aliasing. Applies to any signal (sound or vision)

5. Sampling Sampling happens in life time events. Looking at the weather. Cartwheels classic example. More samples mean more data Sometimes we degrade sound and vision so that we don’t have to sample it so often.

6. Quantisation Sampled signal is still an analogue signal. The sample values may be any value within the signal’s range. I.e. an infinite number of values. The computer deals in numbers, but not in infinite numbers.

7. Quantisation We need to decide the smallest amount a change from one number to the next will represent in our signal. The decision is called quantisation If we have a 4 bit number it can only represent 16 different values. 0 -15 An 8 bit number can represent 256 different values. 0 –255

8. Quantisation Generally an ‘n’ bit number can represent 2 n different values. 0 – (2 n -1) So the more accurately we need to represent a signal, the greater the levels of quantisation. Greater levels of quantisation require more bits (data). 8 bits vision,16 - bits sound. 16 bit gives 0 to 65535

9. Coding Once we have our sampled, quantised data, we need to develop some coding scheme to store in the computer. Or transmit. Need not be difficult. 8 – bit video brightness levels vary from 0 – 255, so simple hexadecimal provides a good coding scheme. Sound waves go negative so a bit more difficult. So could use 2’s complement, offset binary or sign bit coding.

10. Two’s complement Turn all the zeros to ones and all the ones to zero and add 1. Gives us a range from: -32767 to 32767 We lose 80 16 The scale –7F 32767 –…. (numbers between) –011 –00 –FF -1 –…… (numbers between) –81 -32767 All negative numbers start with MSB = 1