69 results about "Aspect ratio (image)" patented technology

Methods and apparatuses for nonlinear scaling of video images. To match the aspect ratios of a video image and the target display area, at least one embodiment of the present invention scales the video image according to one or more nonlinear functions along the horizontal direction and / or the vertical direction. In one embodiment, the nonlinear functions are such that the original aspect ratio of the video image is preserved near the center region (or strip) of the image and the image is gradually stretched (or compressed) as it is mapped to the edges. In one example, the scaling is implemented by the texture mapping functionality of OpenGL using graphics hardware. In one embodiment of the present invention, the nonlinear mapping is constructed according to a polynomial mapping; and, the coefficients of the polynomial are adjustable by a user to trade off distortion between the image center and the image edges, giving the user control over the location and the amount of distortion.

An authentication system using a correlator that correlates an input with a reference wherein at least one of the input and reference comprises a phase volume mask having structures, preferably points, that are each less than about six microns in size and can have an aspect ratio (AR) greater than 1:1 so as to produce a phase encoded random pattern having millions of combinations in a mask that is as small as one square millimeter. The random pattern can be convolved with a second pattern, such as a biometric pattern, to produce a phase convolved mask. The correlator preferably is a nonlinear joint transform correlator that can use “chirp” encoding to permit the input to be located in a different plane than the reference. The correlator optically Fourier transforms images of the reference and input that are thereafter nonlinearly transformed and inverse Fourier transformed by a processor to determine the presence or absence of a correlation spike indicative of authenticity. A spatial light modulator (SLM) can be used as an input or reference and preferably is a liquid crystal panel having pixels or elements whose phase or grey scale intensity can be selectively controlled by a processor. The SLM can be used to display a biometric pattern, preferably scanned in real time from a person, that is correlated against an input or reference that can comprise a label on a card, a tag, or another object.

An area is imaged with a non-standard aspect ratio, e.g. wide aspect ratio, digital imager to produce pixels of image data. At least a portion of the produced pixels of image data corresponding to the captured non-standard aspect ratio image is displayed beginning, for a given frame of pixels of image data, at a column of pixels of image data corresponding to a predetermined column of imaged pixels. The number of columns of display pixels is different than the number of columns of imaged pixels. Motion is detected in the imaged pixels so that a motion responsive portion of the imaged pixels corresponding to the captured non-standard aspect ratio image corresponding to a window of the imaged pixels having motion detected therein can be displayed.

The invention relates to a target detection model training method and device, a computer readable storage medium and computer equipment, and the method comprises the steps: obtaining a feature map ofa sample image during training, and determining an initial detection box in the feature map according to a preset rotation angle, a preset scale and a preset target aspect ratio; adjusting the position of each initial detection frame to obtain the position information of the prediction detection frame, and adjusting the network parameters of the regression network according to the position information and the real position information in the annotation information of the sample image; predicting the prediction probability of the target corresponding to each preset category according to the target detection area determined by the position information of the prediction detection frame; and adjusting the network parameters of the classification network according to the real category information in the annotation information of the sample image and the prediction probability, and then obtaining a target detection model used for performing target detection on the image. According to the scheme provided by the invention, the target detection model can identify the rotation angle of the target in the image, and the positioned target detection frame is more accurate.

The invention discloses a video-image-resolution-based vision adaptive method. The method comprises the following specific steps: (1) inputting an original video image and extracting image attention; (2) setting a clipping ratio, and finding an optimal clipping window area, remained by a video, of each frame with greatest total energy; and (3) adjusting the clipping ratio, finding an optical clipping ratio to obtain an optimal clipping window area, clipping the video image, and scaling the video image to preset target resolution. By the method, under the preset target resolution of the video image, main information of the video image can be remained; when the target resolution is much smaller than the original resolution, a main object in the original video image does not become very small; and when the aspect ratio of the target resolution is not the same as that of the original resolution, the length-to-width ratio of the object does not change, and the video image is clipped and scaled according to the optimal clipping window area with the optimal clipping and scaling ratio to provide a good vision effect for a viewer.

The present invention relates to the field of motion estimation in video processing. Specifically, the invention relates to a video-processing method and device for ascertaining motion vectors for a plurality of first pixel blocks forming a currently processed image region of a currently processed image of an image sequence. The invention addresses the problem of the impact of borders between neighboring image regions in region-based motion estimation on the quality of the video output in video applications like picture-rate up conversion. The video-processing device ( 100 ) of the invention comprises a processing unit ( 104 ), which is adapted to perform motion estimation on an image according to a fragmentation of the image into a number of image regions, each image a region containing the pixel blocks shared by a first number of pixel-block lines and a second number of pixel-block columns in accordance with an adjustable value of an aspect ratio of the image region, and to set a different aspect-ratio value for processing a next image of the image sequence, such that the number of image regions per image remains constant. The dynamic change of the aspect ratio of the image regions implemented in the motion estimation device of the invention reduces the impact of the borders between neighboring image regions and thus improves the quality of region-based motion estimation.

Two images are compared to determine how similar they are. First, a process normalizes each image, then horizontal and vertical byte sequences are derived from each image. A similarity formula is used to obtain a similarity value that represents the similarity between the two images. An approximate pattern matching algorithm is used to determine the error distance between the horizontal byte sequences for the images and to determine the error distance between the vertical byte sequences for the images. The error distances and the length of the byte sequences are used to determine the similarity value. Padding is used to make the aspect ratios the same.

A stereoscopic image processing method for extracting a plurality of dots, i.e., a pixel unit, from a plurality of viewpoint images by each viewpoint image. Data of the plurality of dots serving as a pixel unit is extracted from each viewpoint image, and an aggregate of such the data is defined as a pixel group. A dot cluster of 21 dots surrounded by bold lines constitutes one pixel group. An aperture 1 corresponds to each pixel group, and from a specific observing position, dots in three colors of {1; 11; R}, {1; 11; G}, and {1; 11; B} constituting an upper left pixel of a viewpoint 1 are simultaneously observed. An arrangement of pixels in the pixel group is set so that an aspect ratio of a displayed pitch of the pixel group is rendered most approximate to 1:1.

An imaging apparatus, when an image captured by an imaging unit thereof is not displayed on a display unit, can facilitate framing of an object to be imaged. An aspect ratio setting unit sets an aspect ratio imaged by an image sensor. A system control circuit controls, in a first mode, the display unit to display image data captured by an image sensor in the aspect ratio, and in a second mode, not to display the image data. The system control circuit also displays, in the second mode, information from which at least one of a focus detectable position and a metering area within a shooting area in the aspect ratio is identifiable.

A method of defining model characters of a font. The method includes receiving a string of characters, receiving an image that includes an occurrence of the string, identifying objects in the image, determining, for each respective object, which of the objects satisfies first criteria indicating that the respective object likely corresponds to a character in the string, determining, for each respective object satisfying the first criteria, which of the objects satisfies second criteria indicating that the respective object belongs to a sequence of objects likely to correspond to the string, and defining, for each respective object satisfying the second criteria, a model character for each character of the string based upon a corresponding object of the sequence of objects. The first criteria may include aspect ratio criterion, size criterion, or both, and the second criteria may include alignment criterion, spacing criterion contrast criterion, encompassment criterion, or combinations thereof.

A document alteration detection method compares a target image with an original image by comparing character shape features without actually recognizing the characters. Bounding boxes for the characters are generated for both images, each enclosing one or more connected groups of pixels of one character. The bounding boxes in the original and target images are matched into pairs. Addition and deletion of text is detected if a bounding box in one image does not have a matching one in the other image. Each pair of bounding boxes is processed to compare their shape features. The shape features include the Euler numbers of the characters, the aspect ratio of the bounding boxes, the pixel density of the bounding boxes, and the Hausdorff distance between the two characters. The two characters are determined to be the same or different based on the shape feature comparisons.

The invention relates to a monocular real-time distance measurement method based on a target pixel area and an aspect ratio. The method comprises the following four parts: an ssd model is adopted to realize detection processing of a target, and accurate framing is realized; performing parameter calibration on the camera, obtaining internal parameters and image resolution of the camera, and based on the width and height of target detection, solving a pixel area and width-height ratio of a target object; monocular distance measurement at multiple angles and different distances is realized through the relationship between the pixel area and the distance of target detection and the characteristic of the aspect ratio of a specific target pixel; due to instability of target detection, Kalman filtering is introduced in the real-time distance measurement process, and the stability of real-time distance measurement is improved. Aiming at different imaging at different angles and instability ofreal-time target detection, the distance measurement method based on the target pixel area and the aspect ratio is provided, and the the Kalman filtering method is introduced so that the applicabilityof multi-angle distance measurement is improved, and the stability of real-time moving target distance measurement is improved.

The invention discloses a target tracking method and training method based on a matching-regression network, which are applied to the technical field of image processing, and comprises the following steps: inputting a target to-be-tracked sequence comprising a plurality of frames, and performing target tracking on each frame of image in a target search area; outputting a target center position according to the center matching network, obtaining a feature map output by the last convolution layer of the center matching network according to the determined target center position, and taking the feature map as the input of a boundary regression network; and by the boundary regression network, performing central point divergence according to the input feature map and the target center, determining the boundary position of the target center, and outputting the height and width of an anchor box. According to the method, the center position of the target is determined by using the twin network,and a more accurate anchor box with a variable aspect ratio is output by using two layers of LSTM networks in combination with the time sequence feature information of historical frames in the boundary regression network, so that the accuracy and robustness of target tracking in a video sequence are improved.

A non-transient computer readable medium containing program instructions for causing a computer to perform a method of redrawing an image, the image being usable to decorate a mug having given parameters. The program instructions, according to one embodiment of the method, calculate a first aspect ratio of the mug. The user can then select a selection from the initial image based on the first aspect ratio. Once the selection has been performed, the number of pixels of the selection is adjusted to achieve a high resolution in the transformed image. The program creates a modified image zone. The pixels of the modified image zone are given specific coordinates. A calculation is made to select a corresponding pixel from the selection and places the corresponding pixel in the modified image zone to create the modified image.

The invention provides a target tracking method based on a twin network, and the method comprises the steps: 1, reading a to-be-tracked image sequence or a first frame of a to-be-tracked video, and specifying the position of a to-be-tracked target in a box selection mode; 2, recording the position of the to-be-tracked target, and inputting the RGB image of the target into the network to obtain a feature map; 3, reading in the next frame of the image sequence or the video, and selecting a part of images around the target position of the previous frame to be input into the network to obtain a feature map; 4, performing convolution operation on the two feature maps to obtain a new feature map; 5, convolution is performed on the new feature map by using a small convolution kernel, so that a more abstract feature map can be obtained; 6, enabling the final feature map to pass through a full connection layer, and outputting the coordinate of the target in the current frame, the offset of thecenter of the image and the aspect ratio; and 7, drawing the position of the target in the current frame according to the output in the step 6.

Embodiments described herein simplify the recognition of objects in synthetic aperture radar (SAR) imagery. This may be achieved by processing the image in order to make the shadow caused by the object appear more similar to the object. Alternatively, this may be achieved by processing the image in order to make the layover caused by the object appear more similar to the object. Manipulation of the shadow caused by the object and the layover caused by the object may comprise altering the aspect ratio of the image, and in the case of manipulating the shadow caused by the image, may further comprise transforming the image by reflection or rotation. The aspect ratio of the image may be altered based on information about the image collection geometry, obtained by the SAR.

Provided is an image correction method, acquiring a first aspect ratio of a quadrangle of an image to be corrected and a geometric feature of the quadrangle; acquiring a compensation factor of the quadrangle according to the geometric feature of the quadrangle; calculating a second aspect ratio of the quadrangle according to the first aspect ratio of the quadrangle and the compensation factor of the quadrangle; calculating a transformation matrix according to the second aspect ratio of the quadrangle; and correcting the image to be corrected according to the transformation matrix. By means ofthe solution provided in the present application, an estimated value closer to the original aspect ratio of a captured subject can be obtained, thereby obtaining a corrected image that is substantially consistent with the original aspect ratio of the captured subject.

The invention relates to a method for identifying sheltered and overlapped fruits for a picking robot, which belongs to the field of image identification, proposes a Dense-TRH-YOLO model, fuses a Denseblock module into a backbone network on the basis of YOLOv5, creates a segment path from an early stage layer to a later stage layer, and fuses a Transfomer module into the model, thereby improving semantic distinguishability and reducing category confusion. The method comprises the following steps of: firstly, extracting image features of each layer through a Unit + +-PAN neck structure, and finally, replacing a CIOU of an original model with an Officient IOU Loss loss function to carry out frame regression so as to output a detection frame position and classification confidence, respectively calculating a width-height difference value on the basis of the CIOU so as to replace an aspect ratio, and introducing Focal Loss to solve the problem of difficult and easy sample imbalance.

The invention discloses a tumble behavior identification method based on posture comprehensive characteristics. Comprising the following steps: video data acquisition, human body skeleton diagram extraction, skeleton image data preprocessing, binary image enclosing rectangle construction, LSTM network unit processing, human body effective area calculation, aspect ratio calculation, distance calculation from the human body mass center to the ground, height change rate calculation and tumble occurrence. On the basis of accurately extracting human body skeleton information, the tumble behavior is comprehensively judged by effectively combining various posture characteristics of a human body, so that the detection accuracy is improved; the lightweight Openpose is used, so that the data volume of the lightweight Openpose is relatively light; secondly, an enhanced gradient descent method and an SGD common gradient descent method are used, and the accuracy can be guaranteed while the convergence speed of the model is increased; the trained optimal model can be conveniently deployed to each edge end; the invention is applied to places such as homes and public places.

The invention relates to the technical field of machine vision, in particular to an image recognition monitoring method for defects of small target equipment. The multi-scale small target equipment defect identification monitoring method is characterized by comprising the following steps: (1) constructing a single target detector for a plurality of categories; (2) predicting category scores and position offsets of a group of default bounding boxes fixed on the feature map by using a small convolution filter; and (3) generating predictions of different scales from the feature maps of different scales, and clearly separating the predictions through an aspect ratio. According to the method provided by the invention, an object region is predicted on feature maps of different convolution layers, discretized multi-scale and multi-proportion default boxes coordinates are output, and meanwhile, frame coordinate compensation of a series of candidate frames and confidence of each category are predicted by utilizing a small convolution kernel.

For monitoring an image transformation such as aspect ratio conversion, an image feature is defined by identifying a position in the image having a local spatial maximum value and then identifying four other positions in the image having local spatial minimum values such that the four minimum value positions surround the position of the maximum, a first pair of the minimums lie on a first line passing through maximum and a second pair of the minimums lie on a second line passing through the maximum.

The embodiment of the invention discloses an image processing method and device and a computer readable storage medium. According to the embodiment of the invention, the method comprises: obtaining a to-be-processed image and size information corresponding to the to-be-processed image; determining a target center point according to the position information of the face feature information in the to-be-processed image; determining cutting size information according to an aspect ratio relationship between the size information and the target size information; cutting the to-be-processed image based on the target center point and the cutting size information to obtain a target image; and zooming the target image to the target size information. Therefore, the target center point of the image is determined according to the face feature information, the cutting size information is determined according to the aspect ratio relation between the size information and the target size information for cutting, and the target image with the aspect ratio meeting the requirement is obtained; and the target image is zoomed to the target size information, the thumbnail meeting the requirement is rapidly generated, manual interference is not needed, and the image processing efficiency is greatly improved.

A photography system uses electronically-readable software to process graphic objects that have been captured by a digital camera or scanned to an electronically-readable format for display on a square display screen. The images can be organized according to the aspect ratio, chronologically or according to other criteria selected by a user. The system operates to organize the displayed images in inform rows and / or columns on the page of the display screen and provides consistent per aspect ratio image size for horizontal and vertical images and unrestricted on-screen image placement.

A system and method allows a light colored image of an object such as a check to be detected and captured on a light colored background for uploading to a server for processing. Detection involves detecting edges of objects on the image, drawing a rectangle around the detected edges, testing for an aspect ratio of the rectangle within an approved range, testing for the rectangle being outside of a margin of the image and being a certain percentage of the image, and testing for blur within a tolerable range.

The invention discloses a method for generating anchor points based on geometric attributes, which belongs to the field of deep learning. The method steps include: step 1, extracting the coordinates of all marked bounding boxes in the image, and normalizing the width and height of the marked bounding boxes ; Step 2, clustering the aspect ratio of the marked bounding box by Euclidean distance; Step 3, clustering the size of the marked bounding box by SIoU distance; Step 4, taking the size and aspect ratio cluster center as the interval points, anchor points that generate size and aspect ratio vectors. The invention clusters the aspect ratio and size of the marked bounding boxes by normalizing the marked bounding boxes, and the generated anchor points contain the geometric attributes of the target, and can extract region suggestions to improve the accuracy of target detection.

The invention relates to a fast deep learning remote sensing image target detection method based on candidate region screening, which belongs to the field of remote sensing image processing and includes the following steps: 1) using target training samples to train a deep convolutional network model; 2) inputting target geometry Feature conditions, including target length, width, and aspect ratio range; 3) Analyzing image parameters, mainly resolution information in two directions; 4) Converting target geometric features into pixel limiting conditions according to resolution information; 5) Using The remote sensing image target segmentation technology realizes the target candidate area extraction; 6) according to the target geometric information size limitation condition in step 4, the target candidate area in step 5 is screened; 7) the target candidate area after screening obtained in step 6 Perform feature recognition to determine whether it is a target. The invention proposes to use the target geometric feature information to screen the candidate areas, which improves the target detection efficiency and has a certain inhibitory effect on false alarms.

Embodiments of the present disclosure provide an image processing method, device, electronic equipment, and storage medium. The method includes: collecting an image sequence corresponding to the shooting object; obtaining a reference image in the image sequence; establishing a three-dimensional model based on the reference image and several frames of consecutive images in the image sequence that are continuous with the reference image; According to the three-dimensional model, determine the aspect ratio of the photographed object; perform projection correction on the reference image according to the aspect ratio, and obtain a corrected image corresponding to the photographed object. The present disclosure can avoid the situation that the image is distorted.