Frame selection based on modified entropy and object motion for secured data hiding in videos / Muhammad Fuad Bin Abdullah

Faculty of Computing

Thesis (Master of Science) -- Universiti Malaysia Pahang – 2022

Includes bibliographical references

The growth of internet technology makes hiding information techniques become popular communication. Video has been widely used as media for concealing communication existence by hiding messages. Videos are often compressed to reduce storage and transmission payload at the expense of lower quality due to bandwidth-related issues. Videos provide additional hidden-space in the video data. Compression methods have been widely applied in multimedia applications. The hiding information based on least significant bits will be lost due to compression. The existing hiding messages focus on the imperceptibility and payload of hidden messages. In practice, the hidden message does not require a large amount of bits. Thus, it is vital to develop a hiding information technique that requires a limited payload of hidden messages but it can be resistant against compression. This research proposed a frame selection technique based on object motion and modified entropy in video steganography. The object motions in the video frame were determined by horizontal and vertical motion vectors. The video frames that had object motion were computed by modified entropy. The lowest modified entropy frames were selected for concealing a secret message. The proposed scheme embedded data along with the object motion by modifying Discrete Cosine Transform (DCT) coefficients in the video frames. Six DCT coefficients were selected in the middle frequency using DCT-psychovisual effects of hiding messages. A message was embedded by modifying middle DCT coefficients using the proposed algorithm. The middle frequencies had a large hiding capacity and it relatively did not any give significant effect to the video reconstruction. The proposed frame selections provided additional security where the hidden data were difficult to be detected. The proposed video steganography achieved less perceptual distortion to the human eyes. The experimental results showed that the scheme achieved good robustness of message recovery in terms of Bit Error Rate (BER) and Normalised Cross-Correlation (NC). The experimental results showed that the proposed scheme improved SSIM value of about 1.21% than other existing schemes. The average PSNR and SSIM values for fourteen videos are about 49.67 dB and 0.989, respectively. The recovered message of the proposed steganography scheme can be resistant under video compression. The proposed scheme also achieved good robustness of message recovery under H.264 compression in terms of NC and BER. The average NC and BER values of message recovery under H.264 compression are about 0.95 and 0.2 respectively.