A zero-watermarking registration and detection method for HEVC video streaming against requantization transcoding is provided. To increase an attack-resistance of a registration watermarking, the registration method firstly processes depth values corresponding to respective brightness blocks in a target video streaming with a depth binarization during constructing registration watermarking information through depth features, because the depth binarization well reflects a robustness of the registration watermarking. A first watermarking information matrix including a part of the depth values after the depth binarization is encrypted with a random matrix, so as to increase a safety of the registration watermarking. The registration method directly generates zero-watermarking through the depth features of the video streaming without modifying original carrier information and affecting a watermarking transparency. Meanwhile, the registration method has a strong robustness against attacks, such as the requantization transcoding of quantization parameters within a certain range of variation and common signal processing.

A method for determining a set of modifiable elements in a group of pictures of a coded bit-stream representative of an audio video content is disclosed. The method comprises determining a set of a candidate modifiable elements wherein a candidate modifiable element comprises a modified value of the coded bit-stream and a spatial propagation map associated with the modified value, a spatial propagation map comprising pixels whose decoding is impacted when the modified value is placed in the coded bit stream; determining a heat map for each reference frame, the heat map comprising, for each pixel of the reference frame, an information representative of the usage of said pixel for temporal prediction during the decoding of said part of the coded bit-stream coding a group of pictures; obtaining a set of modifiable elements among candidate modifiable elements, a modifiable element having a spatial propagation map that does not overlap with corresponding heat map.

A method for picture-based barcode encoding and decoding is provided herein. The method for picture-based barcode encoding includes: transforming an original data into an original data bitstream; performing an error correction on the original data bitstream for translating the original data bitstream into an error corrected bitstream; selecting all or part of the picture as an encoded area; calculating a data storage capacity of the encoded area; adjusting a size of the error corrected bitstream or a size of the encoded area for equalizing a data storage capacity of an encoded data bitstream and the data storage capacity of the encoded area; and adjusting a pixel value of the encoded area according to the encoded data bitstream.

Techniques for latent space based steganographic image generation are described. A processing device, for instance, receives a digital image and a secret that includes a bit string. A pretrained encoder of an autoencoder generates an embedding of the digital image that includes latent code. A secret encoder is trained and utilized to generate an embedding of the secret to act as a latent offset to the latent code. The processing device leverages a pretrained decoder of the autoencoder to generate a steganographic image based on the embedding of the secret and the embedding of the digital image. The steganographic image includes the secret and is visually indiscernible from the digital image. Further, the processing device is configured to recover the secret from the steganographic image, such as by training and leveraging a secret decoder to extract the secret.

A method of signing prediction-coded video data, comprising: obtaining a coded video sequence including at least one I-frame (I), which contains independently decodable image data, and at least one predicted frame (P1, P2, P3, P4), which contains image data decodable by reference to at least one other frame; generating a fingerprint (H.sub.I) of each I-frame; generating a fingerprint (H.sub.P) of each predicted frame by hashing a combination of data derived from the predicted frame and data derived from an I-frame to which the predicted frame refers directly or indirectly, wherein the fingerprint of the predicted frame is independent of any further predicted frame to which the predicted frame refers directly or indirectly; and providing a signature of the video sequence including the generated fingerprints.

Techniques for latent space based steganographic image generation are described. A processing device, for instance, receives a digital image and a secret that includes a bit string. A pretrained encoder of an autoencoder generates an embedding of the digital image that includes latent code. A secret encoder is trained and utilized to generate an embedding of the secret to act as a latent offset to the latent code. The processing device leverages a pretrained decoder of the autoencoder to generate a steganographic image based on the embedding of the secret and the embedding of the digital image. The steganographic image includes the secret and is visually indiscernible from the digital image. Further, the processing device is configured to recover the secret from the steganographic image, such as by training and leveraging a secret decoder to extract the secret.

Techniques are described herein for correlating vector metrics for image embedding processing. Techniques can include receiving an image associated with an article, where the image represents a user-defined feature of the article and providing, to a generative model configured to generate image embeddings. An image embedding may be received from the generative model associated with the image associated with the article representing the transfer of the at least one resource from the first entity to the second entity, the image embedding encoding at least the user-defined feature of the article. A vector distance metric may be determined based at least in part on a vector distance comparison between the image embedding and a prior image embedding and one or more operations of a computer system may be controlled based at least in part on the vector distance metric.

Techniques for latent space based steganographic image generation are described. A processing device, for instance, receives a digital image and a secret that includes a bit string. A pretrained encoder of an autoencoder generates an embedding of the digital image that includes latent code. A secret encoder is trained and utilized to generate an embedding of the secret to act as a latent offset to the latent code. The processing device leverages a pretrained decoder of the autoencoder to generate a steganographic image based on the embedding of the secret and the embedding of the digital image. The steganographic image includes the secret and is visually indiscernible from the digital image. Further, the processing device is configured to recover the secret from the steganographic image, such as by training and leveraging a secret decoder to extract the secret.

A method includes: encoding an image to be encoded based on a pixel height of a watermark image and a vertical search range of an image encoder; acquiring a target image block from the image to be encoded based on information of a target region in which the watermark image is inserted into the image to be encoded; generating, based on the target image block and the watermark image, a watermark image to be encoded having a pixel height no more than a sum of the pixel heights of two slices; encoding the watermark image to be encoded to obtain a watermark slice data stream; and generating, based on the watermark slice data stream and the plurality of slice data streams in the first set mode, a plurality of slice data streams in a second set mode of the image to be encoded.

A computer-implemented method for creating an encoder-decoder system for embedding information in a decorative label. The method includes defining a family of encoder functions and a family of decoder functions. Each encoder function of the family of encoder functions configured to encode information as a respective modification of a decorative label. Each decoder function of the family of decoder functions configured to decode an image of a modified decorative label into respective decoded information. The method includes applying an iterative optimization process to determine an optimized encoder-decoder pair. The optimized encoder-decoder air includes an optimized encoder function and an optimized decoder function. The optimized encoder function is selected by the iterative optimization process from the family of encoder functions and the optimized decoder function is selected by the iterative optimization process from the family of decoder functions.