Issue in generating noise and euclidean distance

Question

(a) How do I generate noise such that the maximum levels of the added noise correspond with 3.9% to 50% of the average absolute deviation.

(b) How do I generate white Gaussian noise and uniform noise?

Answer 1

I can only help with b)

Uniform noise (according to wikipedia) only has a uniform distribution, which should make it white as well): http://en.wikipedia.org/wiki/Image_noise#Quantization_noise_.28uniform_noise.29

uni_noise = rand(100000,1); % uniform noise
hist(uni_noise,500) % uniform distribution
pwelch(uni_noise) % also white noise (even power density)

Gaussian White noise requires an even power density and a normal distribution: http://en.wikipedia.org/wiki/White_noise

gauss_noise = randn(100000,1); % gaussian noise
hist(gauss_noise,500) % gaussian (or normal) distribution (bell curve) + zero mean
pwelch(gauss_noise) % even power density