Digimarc Corporation

Advanced image signal processing is used to generate detectability measures. Such detectability measures can predict robustness of embedded printed objects prior to printing. Other technology is described and claimed as well.

Watermarking media content, in combination with blockchain and distributed storage networks, prevents the proliferation of Deepfake content. Digital watermarks are embedded in the audio and video tracks of video clips of trusted content producers at the time the videos are captured or before they are distributed. The watermarks are detected at the social media network's portals, nodes, and back ends. The embedded watermark imparts a unique identifier to the video, that links it to a blockchain. The watermarks also allow video source tracking, integrity verification, and alteration localization. The watermark detectors can be standalone software applications, or they can be integrated with other applications. They are used to perform three main tasks: (1) they alert the Internet user when he watches an inauthentic news video, so that he may discard it, (2) they prevent a Deepfake content from propagating through the network (3) they perform forensic analysis to help track and remove Deepfake content postings.

A neural-network is trained to produce a visibility model based on just noticeable difference training data. The trained network is used to generate visibility masks to control the visual quality of image processing operations, such as digital watermarking, compression and other image editing operations. The visibility model is refined by iteratively adjusting for over and under predictions of a just noticeable difference threshold via an interactive user interface that pinpoints visual artifacts and captures updates from users, who provide adjustments such that the artifact is below what they perceive to be a just noticeable difference. Image processing, such as digital watermarking is performed in real time to provide the user with feedback to ascertain just noticeable difference levels corresponding to watermark embedding operations.

A steganographic digital watermark signal is decoded from host imagery without requiring a domain transformation for signal synchronization, thereby speeding and simplifying the decoding operation. In time-limited applications, such as in supermarket point-of-sale scanners that attempt watermark decode operations on dozens of video frames every second, the speed improvement allows a greater percentage of each image frame to be analyzed for watermark data. In battery-powered mobile devices, avoidance of repeated domain transformations extends battery life. A great variety of other features and arrangements, including machine learning aspects, are also detailed.

The present disclosure relates generally to signal encoding for value documents. One aspect of the technology relates to authentication of value documents comprising line art patterns. In one example, authentication relies of detection of signals encoded within the line art, and expected frequency domain analysis of the line art. Other combinations are described as well. For example, authentication may rely on detection of signals encoded within encoded and then printed documents, expected frequency domain characteristics encoded signals, presence or absence of halftoning, a color check and/or trained classifier analysis. Other aspects, technology, claims and combinations are described as well.

The present disclosure relates to digital watermarking systems that may use visual masks to optimize watermark embedding in digital imagery. A visual mask provides guidance for adjusting digital watermark signal strength, enabling improved trade-offs between watermark imperceptibility and robustness. Multiple embodiments generate visual masks including: (1) a Perceptual Modeling Candidate approach using contrast masking and texture classification based on standard deviation mapping; (2) a wavelet-based approach using Dual-Tree Complex Wavelet Transform for translation-invariant frequency analysis; (3) artificial intelligence approaches employing convolutional neural networks trained to optimize embedding strength while minimizing perceptual distance metrics such as LPIPS; and (4) LPIPS threshold masking that determines optimal embedding strengths by testing multiple candidate strengths. Visual masks enable content-adaptive digital watermarking that places stronger signals in textured regions while maintaining imperceptibility in flat regions, improving visibility-robustness performance compared to uniform embedding approaches.

A waste stream is analyzed and sorted to segregate different items for recycling. Certain features of the technology improve the accuracy with which waste stream items are diverted to collection repositories. Other features concern adaptation of neural networks in accordance with context information sensed from the waste. Still other features serve to automate and simplify maintenance of machine vision systems used in waste sorting. Yet other aspects of the technology concern marking 2D machine readable code data on items having complex surfaces (e.g., food containers with integral ribbing for structural strength or juice pooling), to mitigate issues that such surfaces can introduce in code reading. Still other aspects of the technology concern prioritizing certain blocks of conveyor belt imagery for analysis. Yet other aspects of the technology concern joint use of near infrared spectroscopy, artificial intelligence, digital watermarking, and/or other techniques, for waste sorting. A variety of further features and arrangements are also detailed.

Watermarking media content, in combination with blockchain and distributed storage networks, prevents the proliferation of Deepfake content. Digital watermarks are embedded in the audio and video tracks of video clips of trusted content producers at the time the videos are captured or before they are distributed. The watermarks are detected at the social media network's portals, nodes, and back ends. The embedded watermark imparts a unique identifier to the video, that links it to a blockchain. The watermarks also allow video source tracking, integrity verification, and alteration localization. The watermark detectors can be standalone software applications, or they can be integrated with other applications. They are used to perform three main tasks: (1) they alert the Internet user when he watches an inauthentic news video, so that he may discard it, (2) they prevent a Deepfake content from propagating through the network (3) they perform forensic analysis to help track and remove Deepfake content postings.

Systems, methods and technology for authenticating multi-layered objects using encoded signals such as digital watermarking. In one implementation, a value document includes a substrate and a removable overlayer, with a security feature comprising a digital watermark signal printed to cover portions of both the substrate and overlayer. The digital watermark signal includes spatially arranged signal elements forming a frame of reference. Tampering is detected by evaluating offset values associated with the signal elements, where misalignment within the frame of reference indicates tampering. In some implementations, the value document includes a first digital watermark on the substrate with synchronization and payload components, a token with a second digital watermark, and a removable sticker with a third digital watermark. Additional security features include split-mark configurations, tamper mark configurations, holographic foils with overprinted digital watermarks, and fragile tamper-deterrent features that deform upon tampering attempts, causing detectable signal distortion. Other combinations are also provided.

Two stages of a convolutional neural network are linked by an interconnect that effects a spatial transposition of array data. The spatial transposition can include rotation, scaling, or translation (e.g., in x- or y-directions). A parameter characterizing the transposition (e.g., a parameter identifying rotation angle) can be learned by the same training process that is also used to learn other network parameters, such as layer coefficients. Additionally, or alternatively, data input to a neural network comprises—for each pixel in a patch of imagery—plural data that each indicates a relationship between the value of the pixel, and the value of a neighboring pixel. Some such neural networks can be trained to indicate the presence of a digital watermark signal in the patch of imagery—or a parameter characterizing such a digital watermark signal. Other features and arrangements are also detailed.