Abstract: A system and method for accessing elements of a table in a digital image of the table, including: obtaining the digital image of the table; finding table elements in the digital image based on digital table properties, wherein the table elements define table cells; calculating coordinates of the table cells in the digital image based on the table elements; and accessing content of a selected table cell in the digital image using the coordinates of the selected table element.

Abstract: Technology for inspection for detecting a defect of a printed matter using machine logic informed by machine learning. Some embodiments of the present invention may include one, or more, of the following features: (i) generates defect datasets; (ii) generates defect libraries; (iii) uses the generated defect libraries for deep learning training; and (iv) uses machine learning to detect defects using computer code (for example, a *.jpg format file) corresponding to an image of a piece of printed matter instead of using a visual image (that is, an image of the type that is created when a person takes a picture using a traditional film camera).

Abstract: A hardware camera may include a camera sensor configured to determine input image data. The hardware camera may also include an image signal processor configured to perform one or more image signal processing operations on the input image data. The hardware camera may also include a neural processing unit configured to determine encoded image data by encoding the input image data with an image data encoder portion of a camera autoencoder. The camera autoencoder may be trained based on training image data collected from the camera sensor and a fingerprint specific to the hardware camera. The hardware camera may also include a camera communication interface configured to transmit the encoded image data to a remote computing system, which may determine decoded image data by decoding the encoded image data via an image data decoder portion of the camera autoencoder.

Abstract: Provided are a color adjustment method, a color adjustment device, an electronic device, and a computer-readable storage medium. The method includes: determining an adjustment area corresponding to an edge pixel, where the adjustment area includes multiple pixels; acquiring color information of the edge pixel and color information of each similar pixel in the adjustment area, where a pixel type of the similar pixel is consistent with a pixel type of the edge pixel; and adjusting a parameter value of the edge pixel according to the color information of the each similar pixel, the color information of the edge pixel, and a brightness parameter of the edge pixel.

Abstract: An image processing method for an image processing device according to an embodiment of the present invention comprises the steps of: obtaining a first RGB image by means of an RGB camera; extracting a reflection component from the first RGB image; obtaining a TOF IR image by means of a TOF camera; and obtaining a second RGB image by calculating the reflection component of the first RGB image and the TOF IR image, wherein the TOF IR image is an amplitude or intensity image generated from an IR image with respect to four phases.

Abstract: An approach for fatigue crack detection is described. In one example, a first image of a structure is captured at a first time, and a second image of the structure is captured at a second time. A feature-based image registration is performed to align features of the second image with the first image, and an intensity-based image registration is performed to further align features of the second image with the first image. A registration error map is determined by performing a pixel-by-pixel intensity comparison of the first image and the second image. Additionally, an edge-aware noise reduction process can be performed on the registration error map. The registration error map can be referenced to identify certain fatigue cracks in the structure.

Abstract: In accordance with an aspect of the present disclosure, there is provided a method for acquiring coordinate system conversion information, the method comprising: acquiring three-dimensional information including first lane information corresponding to a lane adjacent to a vehicle, through a LiDAR installed at the vehicle, and a surrounding image including second lane information corresponding to the lane, through a camera installed at the vehicle; acquiring first coordinate system conversion information on the LiDAR and the camera by matching the second lane information with the first lane information; and acquiring second coordinate system conversion information on the vehicle and the camera by using top view image conversion information acquired based on the surrounding image, and the driving direction of the vehicle.

Abstract: An optimal approach for computing convolutions and cross-correlations of large databases of images that can be arbitrarily large. Throughput is maximized by breaking each image into optimal blocks and then using overlap-and-add method to compute the final result. A parallelized 2D FFT is applied over each block that runs a thread for each physical core.

Abstract: An agricultural applicator, that applies material to an agricultural field, includes an on-board, real-time image sensor that senses targets for the material to be applied. A controller controls applicators, such as nozzles or other applicators, to apply the material to the sensed targets.

Abstract: A method of determining volumetric data of a predetermined anatomical feature is described. The method comprising determining volumetric data of one or more anatomical features present in a field of view of a depth sensing camera apparatus, identifying a predetermined anatomical feature as being present in the field of view of the depth sensing camera apparatus, associating the volumetric data of one of the one or more anatomical features with the identified predetermined anatomical feature, and outputting the volumetric data of the predetermined anatomical feature. An apparatus is also described.

Abstract: A method for determining the reflectance behavior of a surface of an object includes: providing a plurality of images of the object which differ in terms of a recording direction and/or in terms of an illumination direction during the recording; creating data sets with entries which each describe a reflectance value derived from the images and an assigned recording direction and an assigned illumination direction, with each data set being assigned to a point on the surface of the object; and determining missing entries in the data sets on the basis of entries of already created data sets.

Abstract: Described herein is a detector for determining a position of an object. The detector includes a sensor element having a matrix of optical sensors, wherein the sensor element is configured to determine a reflection image. The detector also includes an evaluation device configured to select a reflection feature of the reflection image at a first image position in the reflection image, determine a longitudinal coordinate z of the selected reflection feature by optimizing a blurring function fa, and determine a reference feature in a reference image at a second image position in the reference image corresponding to the reflection feature. The reference image and the reflection image are determined at two different spatial configurations, wherein the spatial configurations differ by a relative spatial constellation, wherein the evaluation device is configured to determine the relative spatial constellation from the longitudinal coordinate z and the first and the second image positions.

Abstract: Apparatus for processing image data associated with at least one input image, including a convolutional neural network, CNN,-based encoder configured to provide a plurality of hierarchical feature maps based on the image data, a decoder configured to provide output data based on the plurality of feature maps, wherein the decoder includes a convolutional long short-term memory, Conv-LSTM, module configured to sequentially process at least some of the plurality of hierarchical feature maps.

Abstract: An ultrasound imaging system has an improved dynamic range control which enables a user to reduce image haze with substantially no effect on bright tissue in the image and without increasing speckle variance. A dynamic range processor processes an input image to produce an approximation image which is a spatially low pass filtered version of the input image. The dynamic range of the approximation image is compressed, and image detail, unaffected by the compression, is added back to produce a dynamically compressed image for display.

Abstract: Embodiments of this application disclose methods, systems, and devices for medical image analysis and medical video stream processing. In one aspect, a method comprises extracting video frames from a medical image video stream that includes at least two pathological-section-based video frames. The method also comprises identifying single-frame image features in the video frames, mapping the single-frame image features into single-frame diagnostic classification results, and performing a classification mapping based on a video stream feature sequence that comprises the single-frame image features. The classification mapping comprises performing a convolution operation on the video stream feature sequence through a preset convolutional layer, obtaining a convolution result in accordance with the convolution operation, and performing fully connected mapping on the convolution result through a preset fully connected layer.

Abstract: Augmented Denoising Diffusion Implicit Models (“DDIMs”) using a latent trajectory optimization process can be used for image generation and manipulation using text input and one or more source images to create an output image. Noise bias and textual bias inherent in the model representing the image and text input is corrected by correcting trajectories previously determined by the model at each step of a diffusion inversion process by iterating multiple starts the trajectories to find determine augmented trajectories that minimizes loss at each step. The trajectories can be used to determine an augmented noise vector, enabling use of an augmented DDIM and resulting in more accurate, stable, and responsive text-based image manipulation.

Abstract: The positions of the tools in an automatic milking arrangement are determined by registering, via a camera at an origin location, three-dimensional image data representing the tools whose positions are to be determined. Using an algorithm involving matching the image data against reference data, tool candidates are identified in the three-dimensional image data. A respective position is calculated for the tools based on the origin location and data expressing respective distances from the origin location to each of the identified tool candidates. It is presumed that the tools are arranged according to a spatially even distribution relative to one another. Therefore, any tool candidate is disregarded, which is detected at such a position that the position for the candidate deviates from the spatially even distribution.

Abstract: Provided are a video cropping method and apparatus, a device, and a storage medium. The method includes: obtaining at least one detection box of a first image frame; determining, based on at least one of an importance score, a coverage area, or a smoothing distance of any detection box in the at least one detection box, a cost of the detection box; determining a first detection box having a minimum cost among the at least one detection box as a cropping box; and cropping the first image frame based on the cropping box. Based on a cost of each detection box, the first detection box having the minimum cost among the at least one detection box is determined as the cropping box to crop the first image frame, which can not only improve flexibility of video cropping, but also improve a cropping effect while simplifying the video cropping process.

Abstract: A method for morphological identification, trajectory tracking, and velocity measurement of high-concentration microbubbles, includes collecting images of a microbubble group, performing image pre-processing, segmenting overlapped pixel blocks of the image of the microbubbles, identifying the image of the microbubbles, obtaining morphological information of each of the microbubbles, obtaining an equivalent diameter of each of the microbubbles, calculating a size distribution of the microbubbles; correlating the images of the microbubbles at two adjacent instants, and obtaining a velocity field and motion trajectory of the microbubbles. The method applies to morphological identification and correlation of high-concentration microbubbles and obtains the particle size distribution and velocity field information of the microbubbles.

Abstract: An image processing system (IPS), comprising: an input interface (IN) for receiving a request to visualize image data captured of an anatomy of interest by an imaging apparatus (IA). An energy value determiner (EVD) is configured to determine based on at least one of the image data, the different image data or contextual data, an energy value for forming, from the image data, a monochromatic image. The determining by the energy value determiner (EVD) is based on an energy curve fitted to the image data. the image data forms part of a series of sectional images acquired of the anatomy of interest, or such sectional images derivable from the image data. The sectional images relate to different locations (z) of the anatomy. The energy curve is fitted to energy value control points assigned to at least a sub-set of the different locations (z). Each energy value control point represents a respective known energy value for a respective one of the sub-set of different locations.