The present disclosure relates to a pupil positioning method. The pupil positioning method may include: obtaining an eye image under illumination of a light source; determining a first internal point in a pupil of the eye image; calculating gradient changes of pixel points along a straight line starting from the first internal point toward outside of the pupil; determining a plurality of edge points at an edge of the pupil based on the gradient changes of the pixel points along the straight line; and performing ellipse fitting on the edge points to obtain a pupil center.

Example image processing methods, apparatus/systems and articles of manufacture are disclosed herein. An example apparatus includes an image recognition application to identify matches between stored patterns and objects detected in a shelf image, where the shelf image has a shelf image scale estimate. The example apparatus further includes a scale corrector to calculate deviation values between sizes of (A) a first set of the objects detected in the shelf image and (B) a first set of the stored patterns matched with the first set of the objects and reduce an error of the shelf image scale estimate by calculating a scale correction value for the shelf image scale estimate based on the deviation values.

Systems and methods are disclosed for recommending products or services by receiving a three-dimensional (3D) model of one or more products; performing motion tracking and understanding an environment with points or planes and estimating light or color in the environment; and projecting the product in the environment.

Devices, systems, computer program products and computer implemented methods are provided for dynamically assessing a moving object from a sequence of consecutive volumetric image frames of such object, which images are timely separated by a certain time interval, by: identifying in at least one image of the sequence the object of interest; segmenting the object to identify object contour; propagating the object contour as identified to other images of the sequence; and performing dynamic analysis of the object based on the object contour as propagated.

Methods, systems, workstations, and computer program products that provide automated implant analysis using first and second sets of patient image stacks of a patient having at least one metallic implant coupled to bone. Relevant image stack pairs are selected from the first and second patient image stacks, the image stack pairs having at least one common target object or part of a target object for analysis therein. Bone and the at least one metallic implant are segmented in the first and second image stacks to define segmented objects and/or segmented parts of objects. Selected relevant image stack pairs from the first and second patient image stacks can be registered using the selected segmented objects and/or the segmented parts of objects. Measurements of movement of the implant and/or coupled bone after the registration can be calculated using the selected segmented objects and/or the segmented parts of objects.

In order to ensure that similarity is detected with high accuracy, regardless of the types of images constituting a group of images to be evaluated for similarity, the image processing device includes difference calculation means 11 for deforming one of two or more images constituting an image group in one or more deforming ways, and calculating a degree of difference between the deformed image and the other image or images in the image group for each pixel using multiple ways for similarity evaluation, normalization means 12 for normalizing each degree of difference by each of the multiple ways for similarity evaluation, and difference integration means 15 for integrating normalized degrees of difference.

Device, system, and method of generating a reduced-size volumetric dataset. A method includes receiving a plurality of three-dimensional volumetric datasets that correspond to a particular object; and generating, from that plurality of three-dimensional volumetric datasets, a single uniform mesh dataset that corresponds to that particular object. The size of that single uniform mesh dataset is less than 1/4 of the aggregate size of the plurality of three-dimensional volumetric datasets. The resulting uniform mesh is temporally coherent, and can be used for animating that object, as well as for introducing modifications to that object or to clothing or garments worn by that object.

An image identification method is provided, including: storing at least one normal state image of at least one test object; an automatic codec receiving the at least one normal state image to become a trained automatic codec; at least one camera device capturing at least one state image of the at least one test object; a computer device receiving the at least one state image, and the trained automatic codec performing feature extraction and reconstruction on the at least one state image to generate at least one reconstructed state image; and the computer device comparing the at least one state image and the at least one reconstructed state image, and determining whether the at least one state image is a normal state image. The present invention also provides an image identification system.

Device, system, and method of generating a reduced-size volumetric dataset. A method includes receiving a plurality of three-dimensional volumetric datasets that correspond to a particular object; and generating, from that plurality of three-dimensional volumetric datasets, a single uniform mesh dataset that corresponds to that particular object. The size of that single uniform mesh dataset is less than ¼ of the aggregate size of the plurality of three-dimensional volumetric datasets. The resulting uniform mesh is temporally coherent, and can be used for animating that object, as well as for introducing modifications to that object or to clothing or garments worn by that object.

Systems and methods of improving alignment in dense prediction neural networks are disclosed. A method includes identifying, at a computing system, an input data set and a label data set with one or more first parts of the input data set corresponding to a label. The computing system processes the input data set using a neural network to generate a predicted label data set that identifies one or more second parts of the input data set predicted to correspond to the label. The computing system determines an alignment result using the predicted label data set and the label data set and a transformation of the one or more first parts, including a shift, rotation, scaling, and/or deformation, based on the alignment result. The computing system computes a loss score using the transformation, label data and the predicted label data set and updates the neural network based on the loss score.