Coverless Steganography Based on Motion Analysis of Video

Algorithm 5

Improved information extraction.

	Input: video database , frame averaging window length avg_L, smoothing window size for optical flow calculation W, number of LK pyramid level L, bin number N, mapped index set , and number of index items m
	Output: secret information bitstream
(1)	For i = 1 : m
(2)	Get video ID, frame ID, and subblock ID from the index item i
(3)	P = VideoToFrame (V_video ID)
(4)	For j = 1: FrameNum-avg_L + 1
(5)	Initialize R_sum as all zero
(6)	For k = 0: avg_L − 1
(7)	Convert RGB to gray: orig_Rj+k = Rgb2gray (Pj+k)
(8)	R_sum = R_sum + orig_Rj+k
(9)	End for
(10)	R j = R_sum/avg_L
(11)	End for
(12)	Median filtering: M1 = medfilt (RFrame ID), M2 = medfilt (RFrame ID+1)
(13)	Repeat step 6–8 of Algorithm 3.
(14)	End for
(15)	Connect all the segments as: {Hs1, Hs2, …, Hsm}
(16)	Remove padding bits and the secret information bitstream is recovered: