Automatic Lane Segmentation in TLC Images Using the Continuous Wavelet Transform

Algorithm 1

Subtle lane recovery.

(1) Compute profile derivative, 𝐷

(2) Locate lane boundaries (𝑛 = number of lanes in the current set)

  (a)   Right borders, 𝑅_𝑖; 𝑖=0,⋯,𝑛;𝑅₀ =1

  (b)   Left borders, 𝐿_𝑖; 𝑖=1,⋯,𝑛+1; 𝐿_𝑛+1= image width

(3) Estimate  1𝑚𝑙𝑤=𝑛^𝑛_𝑖∑₌₁𝑅_𝑖−𝐿_𝑖

(4) Estimate  1𝑚𝑙𝑎=𝑛^𝑛_𝑖∑₌₁𝐷(𝑅_𝑖)−𝐷(𝐿_𝑖)

(5) For 𝑖=1,⋯,𝑛+1

    (a)  If 𝐿_𝑖−𝑅_𝑖−1>𝑚𝑙𝑤

    (i)       Compute local maxima positions, 𝑀_𝑗

    (ii)     Compute local minima positions, 𝑚_𝑘

    (iii)    Create the pairs (𝑀_𝑗,𝑚_𝑘), 𝑚_𝑘>𝑀_𝑗

    (iv)    For each pair

        If 0.6  𝑚𝑙𝑤<𝑚_𝑘−𝑀_𝑗< 1.4  𝑚𝑙𝑤 and 𝐷𝑚_𝑘−𝐷𝑀_𝑗>0.3  𝑚𝑙𝑎

        add the new lane to the set.