289 results about "Image scale" patented technology

The invention discloses an optical flow speed estimation method and system based on image scale invariance, and relates to the technical field of image processing, and the method comprises the steps: collecting image data through a camera, carrying out the preprocessing, constructing a pyramid according to the preprocessed image data, and generating a descriptor; calculating a matching error based on the generated descriptor, constructing an optical flow estimation model through the matching error, and then outputting a preliminary result; according to the method, the image pyramid structure and the multi-layer descriptor generation technology are combined, the problems that an existing optical flow estimation method is insufficient in precision and poor in robustness in a dynamic scene are solved, in the generation process of the pyramid structure, the robustness of the optical flow estimation method is improved, and the robustness of the optical flow estimation method is improved. According to the method, a weighting method based on feature complexity and edge response intensity is adopted, the image feature extraction precision under multiple scales is effectively improved, and accurate estimation of the optical flow speed on different scales is ensured.

Techniques are described for using computing devices to perform automated operations for analyzing visual data of images acquired at a building to determine building information that includes building dimensions. The automated determination of building dimensions and other building information may include determining estimated camera height for one or more camera devices while acquiring the images based on identified visible structural building objects of defined types, using the determined image scale information to further determine resulting building dimensions, and associating the building dimension data with a floor plan generated from analysis of the images. Information about such determined buildings may be used in various automated manners, including for controlling device navigation (e.g., autonomous vehicles), for display on client devices in corresponding graphical user interfaces, for further analysis to identify shared and / or aggregate characteristics, etc.

The invention provides a high-precision lightweight steel surface defect detection method based on YOLOv5 improvement, and the method comprises the steps: firstly adding ASF-YOLO design to a Neck part of an original model YOLOv5, and adding an SSFF module, a TFE module and a CPAM attention mechanism module; the SSFF module combines global or high-level semantic information from a plurality of image scales and is used for combined harvesting of combined steel surface multi-scale features; the TFE module is used for capturing local detail features of small objects in the steel surface detection process; the CPAM module is used for integrating information processed by the SSFF module and the TFE module; secondly, a C3 module in an original model YOLOv5 is replaced with a RepNCSPELAN4 module, so that the reasoning speed of the model is increased, the capacity of aggregating multi-scale features is improved, and the model has the high parameter utilization rate; and finally, replacing an original loss function with a WIoU loss function to improve the detection capability of the model, reduce the probability of less detection and missing detection, and improve the overall precision of the model. The precision of the model in a steel defect detection task can be improved, and the real-time detection speed can be guaranteed.

The embodiment of the invention discloses a sample state detection method applied to a biological sample and computer equipment. A specific embodiment of the method comprises the following steps: determining state detection configuration information for a target biological sample group; according to the storage configuration information, moving the target biological sample group into a sample storage device so as to store the target biological samples; in response to the arrival of the sample detection time point, executing the following processing steps on the target biological sample: sucking the target biological sample to a slide manufacturing device through a pipetting device to generate a to-be-detected slide; controlling the electron microscope to collect electron microscope images of the to-be-detected slide under different image scales to obtain an electron microscope image group; generating sub-sample detection information for the target biological sample; and generating sample detection information for the target biological sample group according to the obtained sub-sample detection information group. According to the embodiment, the consistency of sample detection is ensured, and meanwhile, the sample detection efficiency is greatly improved.

Methods for multiple instance learning of tissue sample images are described. The methods may comprise, for example, receiving a whole slide image from a needle core biopsy sample from a subject; identifying a tissue region in the whole slide image; selecting a set of image patches from the identified tissue region; resampling the set of image patches at a plurality of image scales to generate a plurality of resampled image patches; generating image representations for the plurality of resampled image patches; extracting feature vectors based on the image representations; providing the feature vectors as input to a trained machine learning model configured to predict a gene alteration state; and outputting the predicted gene alteration state for the needle core biopsy sample for the subject.

Processing image data using deep neural networks is critical to many systems that desire to understand objects and their environment using camera sensors. Image scaling is a fundamental processing task required when managing image data. Although it is possible to scale image data using standard computer or graphics processors it would be highly advantageous in terms of throughput, latency and power consumption to manage image scaling using dedicated neural network hardware. The inventions contained herein provides methods to use existing neural network hardware to preform image scaling functions. Further, the inventions contained herein describe additional circuitry that can be added to neural network hardware to further enhance image scaling capabilities and efficiencies.

The invention discloses a method for calculating the wave load borne by a pile based on machine vision, and the method comprises the steps: collecting and obtaining a water surface image sequence containing the pile, carrying out the image processing of the water surface image sequence, obtaining a flow field image, carrying out the post-processing of the flow field image, and obtaining an optimized flow field image, the flow field is comprehensively monitored through machine vision, and accurate recognition of the flow field where the pile is located is improved; calibration of a preset image scaling is carried out on the optimized flow field image, pixel coordinates in the optimized flow field image are converted into actual coordinates, pile periphery data under a corresponding flow field are obtained based on a particle image velocimetry (PIV) algorithm, and a velocity vector field of fluid particles can be obtained through the particle image velocimetry algorithm (PIV); the flow velocity distribution, vortex and other fluid mechanics parameters of the whole water area are further deduced, pile periphery data under the corresponding flow field can be obtained, then the wave load borne by the pile is obtained based on the Morison equation, and the defect that according to an existing wave load method, due to the fact that the working environment of underwater ocean engineering structures is complex and severe or the safety requirement is high, the wave load is poor is overcome. The problems of low wave load accuracy and high investment cost are solved, and the accuracy and efficiency of the wave load borne by the pile are greatly improved.

The invention relates to the field of computer image processing, and provides a remote sensing multispectral image fusion method and system based on multi-stage feature correction, and the method comprises the steps: obtaining a spectrum mask through a low-resolution multispectral image and a panchromatic image, and covering the panchromatic image through the spectrum mask to obtain a corrected panchromatic image; performing scale feature enhancement on the low-resolution multispectral image, and performing down-sampling and convolution on the corrected panchromatic image to obtain image scale features; obtaining attention block input, and mapping the attention block input to obtain a mapping vector group; performing image correction by using the mapping vector group to obtain a feature correction vector, performing iteration on the feature correction vector to obtain a target feature correction vector, and obtaining a fusion feature matrix; and iterating the fusion feature matrix to obtain target fusion features, and carrying out feature refinement and convolution on the target fusion features layer by layer to obtain a high-resolution multispectral image. According to the invention, a high-fidelity and high-resolution multispectral image can be obtained.

Provided in the present application are a method and apparatus for generating an image adversarial sample, a method and apparatus for training an image processing model, an image processing method and apparatus, and a device and a medium. The method for generating an image adversarial sample comprises: acquiring an original image sample, and acquiring a feature vector map corresponding to the original image sample (S110); performing image scaling processing on the feature vector map according to an image scale of the original image sample, so as to obtain a standard-scale feature map (S120); using an attention mechanism network of a target type to process the standard-scale feature map, so as to obtain an attention influence map (S130); and on the basis of the attention influence map, adding a disturbance to the original image sample, so as to obtain an image adversarial sample corresponding to the original image sample (S140).

The invention relates to the technical field of model automatic training, and discloses a small model automatic training method based on multi-modal large model cognition intelligent driving, and the method comprises the steps: obtaining each training sample generation task received in a target time period and a task execution state of each model driving end; judging whether the model driving group meets a deadline requirement for executing an exclusive sample labeling strategy or not, and selecting the exclusive sample labeling strategy or a shared sample labeling strategy to execute distribution and automatic labeling actions for all multi-modal training data in each training sample generation task, and sending the training sample set obtained by automatic labeling to a model application end to carry out small model training aiming at the corresponding detection scene. Therefore, by performing selective annotation image proportion sharing on the training sample generation task and establishing a dual-behavior judgment system of the model application end and the model driving end, the task processing amount is reduced, and meanwhile, the influence on the annotation data quality and the training model recognition accuracy is reduced as much as possible.

Systems and methods include techniques associated with one or more machine learning systems analyze, compare, and process one or both of model outputs or model inputs. Content verification may include generating content using a first trained machine learning system and then verifying the initial generation using one or more second trained machine learning systems, such as by generating content associated with a prompt, comparing output labels, or comparing output responsive to changing parameters. Additionally, data preparation may include image scaling and batching methods to improve machine learning outputs.

The invention discloses a method for three-dimensional reconstruction and volume quantification of a middle pulp edema area, particularly relates to the technical field of medical image processing, and aims to preprocess an original image and lay a foundation for smooth three-dimensional reconstruction and volume quantification of the middle pulp edema area. According to the method, points in a section contour are extracted from an OCT B-scan image and mapped to correct positions in a three-dimensional space, and then the upper surface and the lower surface of the edema area are presented by using a visualization means, so that three-dimensional reconstruction of the edema area is realized, a parameterized space of the mesoplasm edema area can be obtained, and a foundation is laid for subsequent volume quantification; the method comprises the following steps of: constructing an inter-frame adjacency relation, reasonably filling a blank area by using a symmetrical frame unit, further dividing the symmetrical frame unit into infinitesimal elements in a parameterized space of a medium pulp edema area, calculating the volumes of all the infinitesimal elements, and realizing volume quantification of the edema area by combining an image scale.

The curvilinear polygon image scaling method includes: placing partial derivatives at points on edges of a curvilinear polygon image to obtain a vector gradient field; integrating the vector gradient field to obtain an auxiliary grayscale image; segmenting the edges; modulating, on the auxiliary grayscale image, to grayscale values of edgelets based on the image scaling parameter and vector gradient field and modulating to nearby grayscale values using a smoothing function to obtain an updated auxiliary grayscale image; intercepting the updated image using a threshold and extracting a polygon contour. If differences between edge placement errors of edgelets, that are between the currently extracted and the original curvilinear polygon image, and the image scaling parameter are less than the preset small value, then the currently extracted contour is used as a scaled image; otherwise, modulating iteratively until the difference is less than the preset small value.

The invention discloses a PDF (Portable Document Format) image screening method driven by intelligent analysis, which belongs to the technical field of document processing and comprises the following steps of: intelligently analyzing a plurality of PDF files to obtain a plurality of pieces of image information for format conversion; traversing the plurality of PDF files based on the plurality of bitmap formats for image capture; performing image content identification on the multiple pieces of image information according to the multiple pieces of image proportion data; identifying the plurality of pieces of image information based on a plurality of text data sets, and traversing a plurality of PDF files according to a plurality of image indexes; simulation preselection is carried out based on the target PDF file, full-text search is carried out based on multiple preselected images, preselected feedback information is sent to the remote control terminal, and an image screening instruction is generated to carry out intelligent screening response on the PDF images. According to the method and the device, the problems of low PDF file image processing efficiency and inaccurate information extraction in the prior art are solved, so that a user faces tedious manual screening and processing flows in document management and data analysis.

The invention discloses an image registration method and device, a computer program product and electronic equipment. The method relates to the field of remote sensing image processing or other related fields, and comprises the following steps: acquiring M remote sensing images, and respectively carrying out image scaling processing on the M remote sensing images to obtain M images to be registered; obtaining a copy image of each to-be-registered image, performing feature extraction on the M copy images by using a feature extraction algorithm, and outputting M groups of feature data; performing preliminary registration processing on the M groups of feature data to obtain M initial registration point sequences, and determining M pre-registration images according to the M initial registration point sequences; performing feature extraction on the M pre-registered images to obtain M sets of registered point sequences; and performing multi-scale screening processing on the M sets of registration point sequences to generate a target registration point sequence, and determining a registration image based on the target registration point sequence. According to the image registration method and device, the problem that the registration precision is low when image registration is carried out on the remote sensing image in the prior art is solved.

The invention provides an omni-directional universal AEB method and system based on pure vision, and the method comprises the steps: obtaining the scale change information of a pixel point corresponding to a panoramic image based on the scene flow information of laser point cloud manufacturing, and generating a panoramic image scale data set; constructing a deep learning network model, training the deep learning network model through the panoramic image scale data set, and generating a panoramic image scale prediction model; and inputting the to-be-measured panoramic image data into the panoramic image scale prediction model to obtain predicted scale change information of the panoramic image scale data, and according to the predicted scale change information, based on the perspective view angle, calculating the collision time between the obstacle and the vehicle, and predicting the collision risk. According to the invention, through panoramic perception of the environment, the traveling track of the vehicle can be analyzed in real time, the potential collision risk can be predicted, the method does not depend on any specific obstacle detection algorithm and is not limited to a white list of any obstacle category, and the adaptability and robustness of the system are significantly improved.

The invention relates to the technical field of machine vision detection, in particular to an automobile part surface defect online detection method and system. The method comprises the following steps: acquiring a grayscale image, calculating the size of a structural element for each pixel based on a local Gaussian Laplacian operator response variance, filtering to obtain a substrate image according to the size of the structural element, and differentiating to obtain a texture image. Determining a Gabor scale and a gray-level co-occurrence matrix statistical direction by using the size, and extracting a cooperative direction gray-level co-occurrence matrix feature; and a weight is set based on the size and is subjected to weighted fusion with a multi-scale rotation invariant local binary pattern feature to generate a texture saliency map, and texture defects are judged. On the substrate image, taking the gray value as the height, and determining a neighborhood calculation curvature feature based on the size to detect the substrate defect. According to the scheme, the image scale can be adaptively analyzed, the background texture is effectively inhibited, and therefore different types of tiny defects such as scratches and pits can be reliably detected.

Provided in the embodiments of the present disclosure are a point cloud coarse registration method and apparatus, and a device. The method comprises: using a plurality of acquisition devices to simultaneously perform, on the same horizontal plane, point cloud acquisition on a target object, so as to obtain a plurality of pieces of point cloud data of the target object, the point cloud data obtained by adjacent acquisition devices overlapping partially; respectively performing discretization projection on the plurality of pieces of point cloud data, so as to obtain a bird's eye view corresponding to each piece of point cloud data (S102); respectively performing two-dimensional image scale-invariant local feature detection on the plurality of bird's eye views obtained, so as to obtain feature data corresponding to each bird's eye view (S103); performing feature matching on the plurality of pieces of feature data obtained, so as to obtain a point cloud transformation matrix (S104); and, on the basis of the point cloud transformation matrix, performing coarse registration on the plurality of pieces of point cloud data of the target object (S105).

In a projection exposure method for exposing a substrate arranged in the region of an image plane of a projection lens with at least one image of a pattern arranged in the region of an object plane of the projection lens, using radiation from a wavelength range around a design wavelength < 260 nm, a substrate (SUB) is coated with a relatively thick radiation-sensitive photoresist layer (RS) and exposed using a projection lens (PO). This lens produces a focus at a design focus position (FOCO) at the design wavelength and, for other wavelengths outside the wavelength range, offset focus positions from an axially extended focus area (ΔFOC) around the design focus position (FOCO). The projection lens is designed as a single-waisted or double-waisted system.During an exposure time interval, different wavelengths from the wavelength range around the design wavelength are used such that the axial extent of the focus area is at least as large as the layer thickness. The projection lens is optically corrected for chromatic aberration (CHV) such that, in the region of the design wavelength, the image scale is essentially independent of the wavelength, so that a change in wavelength of 1 picometer results in a maximum variation of 2 nm between the position of an image point corresponding to an object point within an image field.

In various embodiments, a system can: access a failure image on which a first model has inaccurately performed an inferencing task; train, on a set of dummy images, a second model to learn a visual variety of the failure image, based on a loss function having a first term and a second term, the first term quantifying visual content dissimilarities between the set of dummy images and outputs predicted during training by the second model, and the second term quantifying, at a plurality of different image scales, visual variety dissimilarities between the failure image and the outputs predicted during training by the second model; and execute the second model on each of a set of training images on which the first model was trained, thereby yielding a set of first converted training images that exhibit the visual variety of the failure image.

Techniques are described for performing automatic operations using a computing device to analyze visual data of images acquired at a building to determine building information including building dimensions. Automatic determination of building size and other building information may include determining an estimated camera height for one or more camera devices based on an identified visible structure building object of a defined type while acquiring an image; further determining the resulting building size using the determined image scale information; and associating the building size data with a plan view generated from the image analysis. Information about such determined buildings may be used in a variety of automated ways, including for controlling device navigation (e.g., autonomous vehicles), for display on client devices in corresponding graphical user interfaces, for further analysis to identify shared and / or aggregated features, and the like.

The invention discloses an efficient real-time detection method and system for fabric surface flaw detection, and the method comprises the following steps: collecting a fabric surface flaw data set, carrying out the data enhancement of the data set, and dividing the data set; randomly selecting X images in the training set, inputting the X images into the selected backbone network, performing feature extraction, fusing feature maps, and obtaining N prediction feature maps by a multi-scale adaptive sensing detection head; for the N prediction feature maps, generating a bounding box; performing training to obtain a trained network; predicting the verification set to generate all candidate frames, sorting the candidate frames according to confidence scores, and then performing non-maximum suppression (NMS) on the adjusted candidate frames to remove redundant prediction frames to obtain a final prediction frame; and mapping the prediction frame on the scale of the feature map to the scale of the original map according to the proportional relation to realize the positioning of the fabric image defect. According to the method, the defect identification precision of the surface of the fabric image is higher, and the detection speed is higher.

The invention discloses a displacement monitoring method and system based on a movable monocular camera, and particularly relates to the technical field of camera displacement monitoring. The method comprises the following steps: initially and synchronously acquiring images of a reference target and a to-be-measured target, and acquiring an initial pixel size and an initial center coordinate of the reference target and an initial center coordinate of the to-be-measured target; when the pose of the monocular camera changes, the image is collected again, and the pixel size and the center coordinate after the reference target changes are obtained; calculating an image scaling factor, determining an image coordinate transformation parameter, and constructing a dynamic coordinate transformation relation to obtain theoretical pixel coordinates of the target to be measured; and calculating the displacement according to the deviation between the actual pixel coordinate and the theoretical pixel coordinate of the target to be measured. According to the invention, stable calculation of the displacement of the to-be-measured target can be realized under the condition that the pose of the monocular camera changes, and the applicability and measurement consistency of the displacement monitoring process are improved.

Techniques for image scaling based on emotion detection are described. In some embodiments, the techniques include acquiring one or more images of a user, processing the one more images to generate emotion-specific three-dimensional (3D) positions of ears of the user based on a 3D head geometry and an emotion of the user, where the emotion is identified based on the one or more images of the user, and processing one or more audio signals to generate one or more processed audio signals based on the three-dimensional positions of the ears. Further embodiments include systems and non-transitory computer-readable media that perform the steps of the method.

The invention discloses a steel plate surface defect detection method based on a Swindow-Transform network structure and electronic equipment. The method comprises the following steps of: extracting scale features of an image by using each layer of Stage and outputting a feature map; convolution is carried out on the feature map by using different convolution kernels to obtain a scale feature map; carrying out global pooling on the scale feature map to extract a pooling vector, generating a fusion scale weight according to the pooling vector, and carrying out weighted fusion to obtain a fusion feature map; operating the fusion feature map, introducing a fusion scale weight to obtain classification branch features, performing weighted fusion, and inputting the classification branch features into a classification full-connection layer to obtain a classification prediction result; performing global average pooling on the fusion feature map to obtain global feature representation, processing pooling vectors, fusing fusion scale weights to form regression feature representation, and inputting the regression feature representation into a regression full connection layer to obtain a regression prediction result; and performing feature splicing on the classification feature representation and the global feature representation to generate an interaction weight, and weighting a classification prediction result and a regression prediction result to obtain final classification output and final regression output. The problem of surface defect detection is solved.

In various embodiments, a system can: access a failure image on which a first model has inaccurately performed an inferencing task; train, on a set of dummy images, a second model to learn a visual variety of the failure image, based on a loss function having a first term and a second term, the first term quantifying visual content dissimilarities between the set of dummy images and outputs predicted during training by the second model, and the second term quantifying, at a plurality of different image scales, visual variety dissimilarities between the failure image and the outputs predicted during training by the second model; and execute the second model on each of a set of training images on which the first model was trained, thereby yielding a set of first converted training images that exhibit the visual variety of the failure image.

The present invention relates to the field of target detection technology, and provides a method and system for identifying page UI elements based on target detection, the method comprising: obtaining a page image, constructing a data set, and dividing the data set into a training set and a test set; modifying a feature extraction network, a feature fusion network, and a loss function of a YOLOv8s network model to form an OD_YOLOv8s model based on a full-dimensional dynamic convolution kernel; inputting the training set into the OD_YOLOv8s model for training to obtain a target detection model; inputting the test set into the target detection model, and outputting a UI element detection result of a digital large-screen screenshot. According to the solution of the present invention, the present invention can accurately identify page UI elements, effectively improve missed detection caused by variable scales of target images, and effectively improve the stability of image description generation technology to meet the quality requirements of encoding.

A fundus color photo image grading method and apparatus, a computer device, and a storage medium are provided. The method comprises: obtaining an original image, enhancing the original image, and obtaining a target image (S100); performing color processing on the original image and the target image, and respectively obtaining a first processed image and a second processed image (S200); using a pre-trained grading model to perform processing on the first processed image and the second processed image, and obtaining a target grading result (S300). Multi-color space first and second processed images act as model inputs, and prediction is performed by means of fused features on a whole image scale to implement the classification and grading of common fundus color photo diseases, so as to automatically screen fundus color photos with pathological changes, achieve a pre-screening effect, and improve operational efficiency.

The invention discloses an image segmentation method and device, equipment and a storage medium, and relates to the technical field of medical image processing, and the method comprises the steps: obtaining an initial medical image, and carrying out the enhancement of the initial bottom features of the initial medical image through a target neural network, so as to obtain a target medical image; performing down-sampling on the target medical image in the diagonal direction on different image scales so as to obtain target semantic information and target index values corresponding to the target medical image on different hierarchies; and performing weighted fusion on each piece of target semantic information to obtain fused semantic information, performing up-sampling on the fused semantic information based on the target index value to obtain a mask corresponding to the to-be-segmented target, and performing image segmentation based on the mask. According to the image segmentation method, up-down sampling is carried out based on diagonals, and semantic information of different levels is fused for image segmentation, so that loss of fine-grained information is reduced, and the image segmentation precision is improved.

The invention discloses a corn emergence rate calculation method and system based on unmanned aerial vehicle remote sensing images and machine learning, and relates to the field of corn emergence rate calculation. The problems that errors exist when corn seedlings are extracted and recognized through existing unmanned aerial vehicle remote sensing images, pertinence is not high when a common plant seedling rate detection method is used, and the corn seedlings are difficult to recognize at the initial stage are solved. The method comprises the following steps: constructing an aerial image data set; performing wave band registration, general DSM, single orthoimage, mosaic orthoimage and radiation calibration on the aerial image data set, and then completing splicing; performing color transformation on the spliced image to obtain a binary image; extracting corn features in the binary image according to threshold segmentation and image scale scaling transformation; calculating the number of corn seedlings and the emergence rate extracted by the unmanned aerial vehicle according to the seeding quantity; and constructing a corn emergence rate model according to a random forest algorithm and the actual emergence rate, and obtaining the emergence rate according to the corn emergence rate model. The method is applied to the agricultural field.