Abstract: The present disclosure discloses a system and a method. In example implementations, the system and the method can include receiving an image having a first perspective; generating, via a deep neural network, a depth map corresponding to the image having the first perspective; generating, via the deep neural network, a point cloud representation based on the depth map; projecting the point cloud representation onto a point cloud representation corresponding to an image having a second perspective; generating a depth map corresponding to the image having the second perspective; and generating a synthetic image having the second perspective based on the depth map corresponding to the image having the second perspective and a semantic segmentation map corresponding to the image having the first perspective, wherein the second perspective is different from the first perspective.

Abstract: In a method and apparatus for evaluating image segments for a correspondence calculation, an image signal is received from an image sensor and is evaluated in order to determine the extent to which, for a selected region of the image sensor, noise caused by the image sensor is present; and a resultant evaluation parameter, for an image segment of an image acquired by the image sensor, which describes whether the image segment is suitable for a correspondence calculation, is ascertained. The image segment corresponding to the selected region of the image sensor and the resultant evaluation parameter are ascertained based on the determined extent of the noise.

Abstract: A method for locating text in an image is provided. Binarization processing is performed on the image, to set pixel points corresponding to text in the image to a specified color, and a binary image is obtained. A connected domain of pixels of the specified color in the binary image is detected, to obtain a candidate connected region. Connected domain candidate boundaries corresponding to a text region are extracted from the binary image, and the extracting is based on boundaries of the candidate connected region in the binary image and a text arrangement direction. The text region is obtained in the image by using the connected domain candidate boundaries.

Abstract: Aspects of the disclosure provide for mechanisms for identification of table partitions in documents using neural networks. A method of the disclosure includes obtaining a plurality of symbol sequences of a document having at least one table, determining a plurality of vectors representative of symbol sequences having at least one alphanumeric character or a table graphics element, processing the plurality of vectors using a first neural network to obtain a plurality of recalculated vectors, determining an association between a first recalculated vector and a second recalculated vector, wherein the first recalculated vector is representative of an alphanumeric sequence and the second recalculated vector is associated with a table partition, and determining, based on the association between the first recalculated vector and the second recalculated vector, an association between the alphanumeric sequence and the table partition.

Abstract: A method to reduce background noise in a document image. The method includes extracting, from the document image, a connected component corresponding to a background of the document image, generating a histogram of pixel values of the connected component, generating, using a non-linear mapping function based on the histogram, a non-linear probability distribution of the pixel values in the connected component, generating, based at least on a comparison between the non-linear probability distribution and a predetermined threshold, a replacement range of the pixel values, selecting, from the connected component, a pixel having a pixel value within the replacement range, and converting the pixel value of the pixel to a uniform background color.

Abstract: The present principles relates to a method and device for reconstructing an High-Dynamic-Range image represented by one reconstructed High-Dynamic-Range luma component (??) and two reconstructed High-Dynamic-Range chroma components (Û?, (I)) from a Standard-Dynamic-Range image represented by a Standard-Dynamic-Range luma component (y?, y?1) and two Standard-Dynamic-Range chroma components (u?, v?). The method is characterized in that the method comprises: inverse-mapping (22) said Standard-Dynamic-Range luma component (y?, y?1) to obtain said reconstructed High-Dynamic-Range luma component (??); and correcting (33) said two Standard-Dynamic-Range chroma components (u?, v?) to obtain said two reconstructed High-Dynamic-Range chroma components (Û?, (I)) according to said Standard-Dynamic-Range luma component (y?, y?1) and said reconstructed High-Dynamic-Range luma component (??).

Abstract: An image processing device includes circuitry configured to: calculate a displacement of each of a plurality of corresponding feature regions between a reference image and a base image; calculate, as a evaluation score, difference value between displacements of two feature regions adjacent to each other in at least one of the up/down direction, the left/right direction, and the oblique 45° direction; determine an abnormal region on the basis of the score; classify other feature regions excluding the abnormal feature region; calculate a projection conversion matrix by using the displacement of the other feature regions and the result of the classification; calculate a degree of alignment of each pixel of the reference image with respect to each pixel of the base image by using the matrix; and generate a combined image by combining the reference image converted based on the degree of alignment with the base image.

Abstract: A method of recognizing an object includes comparing a three-dimensional point cloud of the object to a three-dimensional candidate from a dataset to determine a first confidence score, and comparing color metrics of a two-dimensional image of the object to a two-dimensional candidate from the dataset to determine a second confidence score. The point cloud includes a color appearance calibrated from a white balance image, and the color appearance of the object is compared with the three-dimensional candidate. The first or second confidence score is selected to determine which of the three-dimensional candidate or the two-dimensional candidate corresponds with the object.

Abstract: A method, performed by a device, of processing an image, may include obtaining a circular image generated by photographing a target space through a fisheye lens; generating metadata including lens shading compensation information for correcting color information of the obtained circular image; and transmitting the obtained circular image and the metadata to a terminal.

Abstract: A system includes an image sensor, an imaging pipeline, and a display device. The image sensor is configured to capture a first frame of pixel data. The imaging pipeline is coupled to the image sensor to receive the first frame of pixel data. The imaging pipeline includes an adaptive noise filter. The adaptive noise filter is configured to filter a pixel based on noise in the pixel. The imaging pipeline is configured to output a second frame of pixel data. The second frame of pixel data includes pixels filtered by the adaptive noise filter. The display device is coupled to the imaging pipeline to receive the second frame of pixel data. The display device being configured to display the second frame of pixel data.

Abstract: The present invention provides an image processing circuit, wherein the image processing circuit comprises a receiving circuit, an image dividing circuit, a first image enlargement circuit, a second image enlargement circuit and an output circuit. In the operations of the image processing circuit, the receiving circuit receives image data, the image dividing circuit divides a pixel value of each pixel of the image data into two parts to generate first image data and second image data, the first image enlargement circuit enlarges the first image data in a linear manner to generate enlarged first image data, the second image enlargement circuit enlarges the second image data in a non-linear manner to generate enlarged second image data, and the output circuit generates an output image according to the enlarged first image data and the enlarged second image data.

Abstract: A CPU acquires bibliographic information of attached documents listed in an application document and bibliographic information given in each of the attached documents. If the application document or any of the attached document linked to the application document are in the form of image data, the CPU recognizes character information from the image data. Once the CPU determines the levels of agreement between bibliographic information in the application document and the corresponding bibliographic information in the individual attached documents, the CPU reports the bibliographic information in the application document, the corresponding bibliographic information in the attached documents, and the levels of agreement between these bibliographic information.

Abstract: Implementations include receiving an input image including a face, processing the input image through a global transformation network to provide a set of global transformation parameters, applying the set of global transformation parameters to the input image to provide a globally transformed image, processing the globally transformed image through a global representation learning network to provide a set of global features, processing the set of global features through a part localization network to provide a set of part localization parameters, applying the set of part localization parameters to the globally transformed image to provide a locally transformed image, processing the locally transformed image through a part representation learning network to provide a set of local features, and outputting a label representing at least one attribute depicted in the input image based on fusing global feature(s) from the set of global features, and local feature(s) from the set of local features.

Abstract: A method of compression of the color data of point clouds is described herein. A palette of colors that best represent the colors existing in the cloud is generated. Clustering is utilized for generating the palette. Once the palette is generated, an index to the palette is found for each point in the cloud. The indexes are coded using an entropy coder afterwards. A decoding process is then able to be used to reconstruct the point clouds.

Abstract: A method includes identifying, by a processor, a gauge dial pattern in a first file based on a gauge dial template in a second file; identifying, by the processor, a needle pattern in the first file; generating, by the processor, a line pattern that approximates the needle pattern; determining, by the processor, a deflection angle of the needle pattern with respect to the gauge dial pattern based on the line pattern; converting, by the processor, the deflection angle into a reading of the gauge dial pattern based on the gauge dial template; and taking, by the processor, an action based on the reading.

Abstract: A system receives images capturing a non-planar surface from different respective angles, where each of the images captures a section of an area on the non-planar surface, the area including textual or graphical information. The system stitches together the images to obtain a stitched image that captures an entirety of the area on the non-planar surface. The system processes the stitched image to extract the textual or graphical information from the area on the non-planar surface. The system then associates the extracted textual or graphical information with one or more information fields in a report template, and generates a report by populating the one or more information fields of the report template with corresponding associated information within the extracted textual or graphical information.

Abstract: An electronic device is provided. The electronic device includes a memory storing parameter sets, each of which includes one weight and bias sets respectively corresponding to n (where, n>1) occlusion levels among a plurality of occlusion levels within a certain range and at least one processor configured to obtain output data by inputting input data to a neural network.

Abstract: A method of updating background model for an image processing device is disclosed. The method includes receiving at least a background image, counting a number of occurrences of a color of each pixel of the background image according to a color information of each pixel of the at least a background image, to establish a background model including a statistical color information about the color and the number of occurrences of the color, receiving a current image, and determining whether to update the number of occurrences of the color in the background model corresponding to a color of each pixel of the current image according to a color information of each pixel of the current image and a result of comparing a random number with a threshold.

Abstract: A resolution conversion of image data of a photographic image is performed, and based on information related to a distance from a focal plane in the photographic image and the resolution conversion, processing to control the sharpness of an image that the image data resulting from the resolution conversion represents is performed.

Abstract: A method includes obtaining a first image of a patient procured during an X-ray, analyzing the first image for one or more unmasked anomalies, shifting the first image to provide a shifted image, obtaining a residual image comprising a combination of the first image and the shifted image, and analyzing the residual image for one or more masked anomalies, where the one or more masked anomalies include anomalies that went undetected in the analysis of the first image for the one or more unmasked anomalies due to a presence of one or more masking features in the first image.