We propose a fast and secure method for compression independent video encryption. A divide-and-conquer type approach is taken where frames in a video are processed at shot level. Temporal redundancy among the frames in a shot is exploited next to derive the difference frames, where some of the pixel values are already zero. Further sparse representations of the pixels are obtained by thresholding the non-zero values. We also propose a new chaotic map, namely, Logistic Tent Infinite Collapse map (LT-ICM) which has better chaotic range, complex behavior, unpredictability, ergodicity, and sensitivity towards initial values and controlling parameters. We permute the pixels of each shot (left after thresholding) with reference to the indices of sorted chaotic sequence generated by LT-ICM. Substitution operation is applied next to change the pixel values. The substituted pixels are arranged into frames and pixels are further substituted at frame level with another chaotic sequence generated by LT-ICM. Experimental results establish that our proposed method is efficient in terms of encryption time, correlation coefficients and entropy values compared to some of the existing methods.

In recent years, video encryption based on chaotic maps has emerged as an important research problem. Properties like high randomness, ergodicity, sensitivity to controlling parameters and initial states make chaotic map based video encryption more popular than the conventional encryption techniques. In this paper, we propose a compression independent method for video encryption at frame level where a new confusion technique, based on sort index permutation, is introduced. During shuffling, a new sequence is generated for every video frame under consideration. Generated pseudorandom chaotic sequence depends on initial values, controlling parameters and the video frames. We have used a logistic sine coupling map (LSCM) for generating highly pseudorandom sequence. Block level diffusion has been employed for scrambling the frames. The proposed framework is more robust against known and chosen plaintext attacks as we have incorporated the size and sum of all pixel values for generating random chaotic sequence. Experimental results clearly demonstrate that our method is effective in terms of security, efficiency and robustness compared to some of the existing approaches.