Given a simple image of size 10 × 10 whose histogram models the symbol…

2014

Given a simple image of size 10 × 10 whose histogram models the symbol probabilities and is given by

p1p2p3p4abcd

The first order estimate of image entropy is maximum when

  1. A.

    \(a = 0, b = 0, c = 0, d = 1\)

  2. B.

    \(a=\frac{1}{2}, b=\frac{1}{2}, c=0, d=0\)

  3. C.

    \(a=\frac{1}{3}, b=\frac{1}{3}, c=\frac{1}{3}, d=0\)

  4. D.

    \(a=\frac{1}{4}, b=\frac{1}{4}, c=\frac{1}{4}, d=\frac{1}{4}\)

Attempted by 118 students.

Show answer & explanation

Correct answer: D

Answer: The entropy is maximized when a = b = c = d = 1/4.

Reasoning: For probabilities a, b, c, d the first-order entropy is H = - (a log a + b log b + c log c + d log d), subject to a + b + c + d = 1.

  • Use a Lagrange multiplier λ to enforce the constraint: consider L = -Σ p_i log p_i + λ(Σ p_i - 1).

  • Differentiate with respect to each probability p_i (where p_i is a, b, c, or d): ∂L/∂p_i = - (log p_i + 1) + λ = 0. This implies log p_i is the same for every i, so each p_i is equal.

  • With four symbols the common value is p_i = 1/4, so a = b = c = d = 1/4.

  • The maximum entropy value is H = -4*(1/4) log(1/4) = log 4. If logs are base 2, this equals 2 bits.

Therefore the first-order estimate of image entropy is maximized when all four symbol probabilities are equal: a = b = c = d = 1/4.

Explore the full course: Nta Ugc Net Paper 1