Detecting forged images and Deepfake videos remains a pressing challenge. Most existing detection methods rely on datasets that contain specific types of tampering. However, as forgery techniques continue to evolve, new forms of manipulation emerge and may undermine these approaches. Furthermore, media content often undergoes compression during dissemination, which can obscure manipulation traces and reduce detection accuracy. This study proposes a forgery detection scheme that leverages block-level consistency. A source-related feature extractor is employed, followed by a Siamese network to classify the compression levels. For image tampering detection, the heatmap is examined to determine manipulated regions. In Deepfake video detection, the proposed scheme focuses on facial regions and evaluates frame-to-frame similarity to assess authenticity. The effectiveness of the proposed approach is verified using publicly available datasets.

A video forensic scheme is proposed in this research to determine if an investigated H.264/AVC video stream has been tampered by such editing operations as shot deletion, insertion or replacement. The scheme is designed based on H.264/AVC de-blocking filter, in which the Boundary Strength (BS) used for evaluating the filter strength of block boundaries is utilized to form two features, VRF (Variation of Residual Footprint) and DOF (Degree of Fragments). The distances between the detected peaks or large values from VRF and DOF are used to acquire the size of the original GOP (Group of Pictures) employed in the encoding before video editing. Possible positions of shot manipulation are identified based on the periodic nature of these peaks. Several tests and a state-of-the-art anti-detection method that modifies coefficients and coding modes are applied in the experiments to verify the feasibility of the proposed scheme.

Assigning an appropriate Quantization Parameter (QP) to the Intra-coded frames is important to the video coding. In this research, a content adaptive Rate-Quantization (R-Q) model is presented to predict the bit usage of Intra-coded frames in H.264/AVC. The relationship between the QP of a macroblock and the block complexity is derived so that a suitable QP can be decided under a target bit-rate. Since the proposed model is built on macroblocks, Region of Interest (ROI) coding can also be achieved. By adjusting the QP value at the macroblock level, more bits can be assigned to the ROI to better maintain its perceptual quality. The experimental results show the feasibility of the proposed R-Q model.

Accurately estimating the resultant quality or distortion associated with quantization parameter (QP) is very helpful to video encoding. In this research, a content-adaptive distortion-quantization model for H.264/AVC is proposed to predict the distortion level, which is defined as the difference between the original video frame and the decoded one in the sum of squared errors. The proposed model has only one adjustable parameter related to the macroblock content and provides a mapping between QP and the corresponding distortion before the exact encoding process. Given a targeted frame quality measured in peak signal to noise ratio (PSNR), this model can help to assign a suitable QP value to each frame. Two applications are then presented, i.e., the single-pass constant frame PSNR coding and the two-pass coding with the additional bitrate or storage constraint, both of which may facilitate such applications of video archiving and editing. The experimental results show that the targeted PSNR of each decoded frame can be achieved effectively by the proposed mechanism.

A feature-based digital image watermarking scheme is proposed to resist geometrical transformations. Such interest point extraction as scale-invariant feature transform is applied to form the invariant regions for signal embedding. Along with the watermark, the pilot signal used for synchronized detection will also be embedded to help to determine larger areas for improving the payload of watermarking and/or increasing the confidence of watermark detection. Both embedded signals are weighted by the just noticeable difference of Watson's perceptual model to meet the requirement of invisibility. In the detection phase, after the same interest point extraction, synchronization can be achieved via adjusting the related affine parameters of grids on the investigated image to detect the possible hidden pilot signal so that the watermark can be successfully retrieved afterward. Experimental results show the feasibility of the proposed method.

An efficient intra-prediction mode decision mechanism for H.264/AVC is presented in this research. A projection-based approach, which employs the reconstructed surrounding pixels and block content to compute the predicted block residuals (PBR), can effectively eliminate less probable modes from the computation of Rate Distortion Optimization. Both the projected vectors and corresponding predictors can be formed by shifting/adding the related data so the proposed scheme is suitable for hardware implementation. According to the PBR and coding information acquired during the encoding process, some prediction modes can also be skipped to further accelerate the intra coding. The experimental results show that the proposed scheme can effectively reduce the encoding time with slight video quality degradation and bit-rate increment.

A content-adaptive Distortion-Quantization (D-Q) model for intra coding of H.264/AVC is proposed in this research. A single-parameter Distortion- Quantization (D-Q) relationship is derived for predicting the coding performance associated with the Quantization Parameter (QP). The content complexity of a macroblock (MB), which is evaluated by using the Sum of Absolute Transformed Difference (SATD), will be employed to determine the model parameter so that a suitable QP can be assigned. In order to acquire a video with more consistent quality, we propose two schemes, i.e. the two-pass method and the single-pass scheme with self adaption. In the two-pass scheme, the resulting frame quality after the first encoding pass using the initial QP is evaluated and a refining factor will be computed to determine a more appropriate QP. In the single-pass adaptive scheme, the adjustment is applied by referencing the model accuracy in the previous frame. The experimental results show that the assigned QP can effectively help to achieve the targeted frame quality in the H.264/AVC intra coding.

A content authentication scheme for digital videos is proposed in this paper. In order to prevent the content from being unnoticeably tampered with by using digital editing facilities, we employ the approach of scalar/vector quantization on the reliable features extracted from video frame blocks to form the authentication code, which is transmitted along with the video. The resulting authentication code is sensitive to malicious modifications of video data but resilient to allowed lossy compression processes, such as H.264/advanced video coding (AVC). The integrity of video content can thus be guaranteed if the extracted feature matches the transmitted authentication code. Experimental results will show the feasibility of the proposed scheme.

A new application of exploiting digital watermarking techniques for annotating traffic surveillance videos is presented in this paper. The information of each vehicle collected from other sensors/sources will be embedded into the corresponding pixels in the recorded video to facilitate data management. The traffic scene captured by a video camera will be analyzed first and the individual vehicles are extracted and tracked by using Kalman filtering for effective watermarking. The scheme is integrated with H.264/AVC, which is assumed to be adopted by the visual surveillance system, to achieve an efficient implementation. The issues of payload, effective embedding/detection and rate/distortion are taken into account to fulfill the requirements of such an application. The experimental results demonstrate the feasibility of this potential system.