The application provides a liveness detection method capable of implementing liveness detection, and a liveness detection system that employs the liveness detection method. The liveness detection method comprises: irradiating an object to be detected with structured light; obtaining first facial image data of the object to be detected under irradiation of the structured light; determining, based on the first facial image data, a detection parameter that indicates a sub-surface scattering intensity of the structured light on a face of the object to be detected; and determining, based on the detection parameter and a predetermined parameter threshold, whether the object to be detected is a living body.

A multimodal remote sensing image matching method and system integrate different local feature descriptors for automatic matching of multimodal remote sensing images. First, a local feature descriptor, such as the Histogram of Oriented Gradient (HOG), the local self-similarity (LSS), or the Speeded-Up Robust Feature (SURF), is extracted for each pixel of an image to form a pixel-wise feature representation map. Then, the three-dimensional Fourier transform (namely 3D FFT) is used to establish a fast matching similarity metric in a frequency domain based on the feature representation map, followed by a template matching scheme to achieve control points (CP) between images. In addition, the novel pixel-wise feature representation technique named channel features of orientated gradients (CFOG), which outperforms the pixel-wise feature representation methods based on the traditional local descriptors (e.g., HOG, LSS and SURF) in both matching performance and computational efficiency.

In an embodiment, a method for removing flare in images is proposed. The method includes utilizing an image capturing unit to continuously take an equally-exposed image and under-exposed images with variant exposure settings; determining whether a flare is in the equally-exposed image; in response to determining that the flare is in the equally-exposed image, aligning image objects of the under-exposed images with respect to the equally-exposed image; and removing the flare in the equally-exposed image by using the aligned under-exposed images, wherein the under-exposed images used to remove the flare in the equally-exposed image are normalized in brightness based on the brightness of the equally-exposed image and the normalized under-exposed images are then denoised. Also, a system and a non-transitory computer-readable medium performing the method are provided.

Systems and methods herein describe a malware visualization system that is configured to access a computer file, generate a first image of the computer file, determine a frequency count of bi-grams in the computer file, compute a discrete cosine transform (DCT) of the frequency count of bi-grams, generate a second image of the computer file based on the DCT of the frequency count of bi-grams, analyze the first image and the second image using an image classification neural network and generate a classification of the computer file.

A multimodal sensing system includes various devices that work together to automatically classify an action. A video camera captures a sequence of digital images. At least one other sensor device captures other sensed data (e.g., motion data). The system will extract video features from the digital images so that each extracted image feature is associated with a time period. It will extract other features from the other sensed data so that each extracted other feature is associated with a time period. The system will fuse a group of the extracted video features and a group of the extracted other features to create a fused feature representation for a time period. It will then analyze the fused feature representation to identify a class, access a data store of classes and actions to identify an action that is associated with the class, and save the identified action to a memory device.

The present invention relates to a method for generating a biometric signature of a subject comprising: obtaining a plurality of sequential video frame images of a moving subject from a video segment; obtaining a portion of each frame comprising a surrounding of the moving subject; carrying out a transformation function to the frequency domain on one or more of said portions of the frames comprising a surrounding of a of the subject; and optionally saving the spectral characteristics of said transformation function in a repository. The present invention also relates to a system for carrying out said method.

Markov random field parameters are identified to use for covariance modeling of correlation between gradient terms of a loss function of the classifier. A subset of images are sampled, from a dataset of images, according to a normal distribution to estimate the gradient terms. Black-box gradient estimation is used to infer values of the parameters of the Markov random field according to the sampling. Fourier basis vectors are generated from the inferred values. An original image is perturbed using the Fourier basis vectors to obtain loss function values. An estimate of a gradient is obtained from the loss function values. An image perturbation is created using the estimated gradient. The image perturbation is added to an original input to generate a candidate adversarial input that maximizes loss in identifying the image by the classifier. The neural network classifier is queried to determine a classifier prediction for the candidate adversarial input.

Disclosed are techniques for determining object throughput. A method may include obtaining first data representing a first image corresponding to a first time, identifying a first portion of the first data that depicts a first object at a first location, obtaining second data representing a second image corresponding to a second time, identifying a second portion of the second data that depicts the first object at a second location, obtaining third data indicating a counting threshold, determining based at least on the third data and the second location, that the first object satisfies the counting threshold, generating a value indicating a number of objects satisfying the counting threshold, the number of objects including the first object, generating a data value indicating a throughput of the number of objects based on the value indicating the number of objects satisfying the counting threshold and elapsed time between the first and second times.

Methods for determining whether a mark is genuine are described. According to various implementations, a computing device (or logic circuitry thereof) receives (e.g., via a camera or via a communication network) an image of a candidate mark (e.g., a one-dimensional or two-dimensional barcode), uses the image to make measurements of a characteristic of a feature of the candidate mark, resulting in a profile for that feature. The computing device filters out, from the feature profile, all spatial frequency components that are indicated to be sibling frequency components. In some embodiments, the computing device carries out the reverse procedure, and filters out all spatial frequency components except for those indicated to be sibling frequency components.

An individual identification system includes: a storing unit for storing an image capture parameter in association with data characterizing a surface of a reference object; an acquiring unit that, when data characterizing a surface of an object to be matched is input, calculates an approximation degree between the input data and each data stored in the storing unit, and acquires the image capture parameter applied to the object to be matched from the storing unit based on the calculated approximation degree; a condition setting unit that sets an image capture condition determined by the acquired image capture parameter; an image capturing unit that acquires an image of the surface of the object to be matched under the set image capture condition; an extracting unit that extracts a feature value from the acquired image; and a matching unit that matches the extracted feature value against a registered feature value.