790 results about "Image synthesis" patented technology

The invention provides an industrial product defect automatic classification method and system, and the method comprises the steps: collecting original industrial product defect image data, and constructing a labeled image sample set and an unlabeled image sample set; constructing a training image sample set based on the labeled image sample set and the unlabeled image sample set in combination with a plurality of image synthesis strategies; based on the training image sample set, introducing a transfer learning strategy and fusing an attention mechanism, and constructing and optimizing an industrial product defect classification model; performing semi-supervised joint training and online learning based on the training image sample set and the real-time small-batch image sample set; and constructing an industrial product defect identification log based on the real-time image flow sample set, the classification model parameters and the corresponding classification prediction function. On the basis of multi-strategy image enhancement and semi-supervised training, transfer learning and a channel attention mechanism are fused, expansion of industrial product defect image samples and fine defect identification are achieved, and the method is suitable for an intelligent defect detection system in various industrial manufacturing fields.

The invention relates to the technical field of three-dimensional reconstruction, in particular to a lightweight three-dimensional reconstruction method and system based on a three-dimensional Gaussian model, and the method comprises the steps: constructing the three-dimensional Gaussian of a target scene based on a multi-view original image of the target scene through employing a 3DGS algorithm; identifying the minimum resolution of each Gaussian primitive in the three-dimensional Gaussian, setting a spherical region according to the minimum resolution, taking the intersection result of the spherical region and the Gaussian primitive as a region evaluation score, and trimming the redundancy region geometry in the three-dimensional Gaussian in combination with the opacity; and taking the maximum zoom scale of the Gaussian primitive as a spatial density proxy variable, and dynamically distributing spherical harmonic coefficient orders according to a local spatial density proxy variable so as to meet the rendering requirements of different three-dimensional Gaussian geometric regions. On the premise that the image synthesis quality is not lost, the model scale and the transmission overhead can be remarkably compressed, the resource constraint condition of the edge computing equipment can be effectively adapted, and the higher training speed and the better rendering effect are achieved.

The invention belongs to the technical field of image processing, and discloses a maskless image synthesis method and system based on a diffusion model, and the method comprises the steps: inputting a background image and a noise image containing a target object into a pre-trained potential diffusion model, and enabling the potential diffusion model to execute a reverse denoising process under the guidance of a condition vector, the target object is synthesized in a self-adaptive mode, and a synthesized image is obtained; wherein the condition vector acquisition mode comprises the following steps: encoding a reference image containing a target object into an object embedding vector, and generating a group of learnable background cue words according to the object embedding vector; splicing with the background prompt word; and performing spatial mapping on the spliced feature representation to obtain a condition vector. The method can overcome the defects that in an existing image synthesis technology, the process is tedious, the efficiency is low, a large amount of manual intervention is needed, the synthesis result lacks the sense of reality and harmony, and especially the foreground object cannot be adaptively adjusted according to the background environment.

The invention discloses a CBCT (Cone Beam Computed Tomography) high-quality CT (Computed Tomography) image synthesis method based on structure prior guidance, which comprises the following construction steps of patient data collection and arrangement, space guidance deformable attention convolution, structure prior guidance type conditional diffusion image synthesis and reasoning process and quality evaluation. The method specifically comprises the steps of technical design of a space guiding module, technical design of a deformable sampling kernel generator, technical design of an attention weight modulator, contour extraction operation, diffusion modeling operation, conditional denoising network design and target training. The method provided by the invention has the advantages of structural perception, high spatial selectivity, high modulability and the like; and the reduction precision of the model on the key anatomical region is enhanced by taking a contour map as a structure priori condition. The whole diffusion generation network realizes unification of local detail reservation and global structure stability while maintaining noise robustness, and gradually generates a high-quality pseudo CT image which is close to a real CT modal and has excellent structure consistency.

The invention relates to the technical field of computer vision and medical image processing, in particular to an intra-intestinal monocular pose estimation method based on diffusion model deformation field prediction, which comprises the following steps of: inputting two frames of enteroscopy images of an intestinal capsule robot with adjacent time sequences, respectively extracting image features of front and back frames of enteroscopy images through a parameter-shared double-path depth map generation network, and generating corresponding depth maps; solving six-degree-of-freedom pose parameters of the intestinal capsule robot by utilizing the pose estimation network; generating a three-dimensional non-rigid deformation field through the non-rigid deformation field prediction network based on the physical regularization diffusion model; reconstructing a composite image aligned with the real previous frame of enteroscope image through an image synthesis module; completing self-supervised training; the problems that in an existing intestinal capsule robot positioning technology, a non-rigid deformation field lacks biophysical constraints, three-dimensional reconstruction geometric distortion is caused by monocular information, and collaborative optimization is lacked in deformation field and pose parameter decoupling are solved.

Apparatuses, systems, and techniques to infer synthesized images including digital representations of objects blended realistically with appropriate background images. In at least one embodiment, background image data and a description of semantic features for a type of object are fused together to infer such a synthesized image using one or more neural networks.

The invention discloses a target detection method and device, equipment, a storage medium and a product, and relates to the technical field of image recognition. And synthesizing the first front view image and the second front view image into a central eye image. Data labeling is performed on the multi-camera data and the central eye image, a labeling data set special for small target detection is constructed, and labeling efficiency and quality are remarkably improved. Performing aerial view coding on multi-scale features contained in the multi-camera data based on a cross-space cross attention mechanism of aerial view query position and image semantic joint modeling to obtain aerial view features; and on the basis of the annotation data set, aligning the aerial view features to the central eye image to obtain aerial view enhancement features, and realizing small target semantic enhancement in the aerial view features. And carrying out target identification on the aerial view enhanced features so as to determine a detection target contained in the multi-camera data. According to the target detection scheme provided by the invention, the detection accuracy and robustness of the small-size target are remarkably improved.

The invention discloses a display screen defect image synthesis method of a hybrid network model, which is applied to the technical field of machine vision and aims to solve the problem that the prior art depends on a large number of defect samples to detect display screen defects and is difficult to adapt to scenes of industrial defect products. The method comprises the following steps: firstly, preprocessing display screen defect image data; carrying out positive sample training learning by adopting an unsupervised network model; constructing an image-level defect synthesis network model, and synthesizing multi-scale and multi-form display screen defect samples to assist in training discriminator parameters of the unsupervised network model; a feature-level defect synthesis network model is constructed, initial defect data is generated by adopting a Gaussian noise model, an initial feature sample is generated by utilizing a self-adaptive gradient rise and refined truncation projection method, and the initial feature sample and a normal feature sample are superposed to synthesize a feature-level defect sample to assist in training discriminator parameters of the unsupervised network model; and finally, the trained unsupervised network model is adopted to carry out detection and segmentation positioning on a to-be-detected image sample.

The invention discloses a concrete defect image generation method based on mask guidance. The method is suitable for data enhancement and structured image synthesis tasks of defect images such as concrete cracks under the small sample condition. According to the method, a space mask mechanism is introduced, a defect area is accurately positioned and extracted from a reference image, and the injection accuracy and control granularity of defect features are effectively improved. The method comprises three main stages: a multi-scale theme-background feature extraction stage, an adaptive time feature generation stage and a multi-scale cross attention image denoising stage. In the feature extraction stage, spatial attention enhancement and fine-grained semantic alignment are carried out on a reference image and a subject text, so that unified representation of multi-modal features is realized; in the time feature generation stage, feature adaptive time is introduced, and time step related weights are generated to dynamically regulate and control the fusion proportion of theme and background features; in the image denoising stage, guidance denoising and image generation are completed in combination with UNet and cross attention.

In some implementations, the techniques described herein relate to a method including: (i) training, by a processor, a machine learning model to create composite images from background scenes and foreground objects, (ii) identifying, by the processor, a digital image file that comprises a background scene and an additional digital image file that comprises a foreground object, (iii) compositing, by the machine learning model executed by the processor, the digital image file that comprises the background scene and the additional digital image file that comprises the foreground object to produce a composite digital image file that comprises the foreground object and the background scene by performing at least one of a channel concatenation step and a reverse diffusion sampling step, and (iv) causing display, by the processor, of the composite image file that comprises the foreground object and the background scene.

The invention relates to a diffusion prior synthesis and optimization method for cross-modal medical image synthesis, which is used for solving the problems of strong dependence on source modal data, high acquisition cost and the like in the existing method. According to the method, source modal data is not needed, training is carried out only based on single target modal data, and a diffusion prior synthesis (DPS) module and a diffusion prior optimization (DPO) module are included. The DPS encodes the image to a potential space guided by general diffusion prior through a probability flow ordinary differential equation, and decodes the image into an initial image through a target diffusion model; the DPO optimizes an initial image through a linear inverse problem, recovers high-frequency details and corrects discrete errors, thereby improving image fidelity and stability.

An embodiment provides a method of producing a three-dimensional (3D) model of an object based on a single, frontal input two-dimensional (2D) image. In one example a method includes obtaining an actual, frontal 2D image of an object and generating a pair of synthetic 2D multiview images of the object based on the actual, frontal 2D image of the object. A 3D model of the object is produced based on at least the pair of synthetic 2D multiview images. The 3D model conserves an asymmetry of the object. An output using the 3D model of the object is produced that conserves the asymmetry.

The invention relates to a few-sample new-view-angle image synthesis method based on multi-scale mixed perception and state space collaboration, and solves the problems of depth ambiguity, excessive smooth texture and serious artifacts in a shielding region caused by insufficient geometric constraints in a generalized neural radiation field in a few-sample scene compared with the prior art. And the problems of high quadratic calculation complexity, large video memory occupation and low rendering efficiency caused by introducing 3D Transform to perform global modeling in the prior art are solved. And the defect of inaccurate geometric surface positioning caused by noise interference in the traditional sampling strategy based on attention weight is overcome. The method comprises the following steps of data set construction and sparse input acquisition; setting a hierarchical bidirectional feature aggregation network; multi-scale features are extracted; carrying out global context modeling; extracting and fusing local geometric features; and generating a new view angle image. According to the method, multi-scale features are extracted through the hierarchical bidirectional feature aggregation network, collaborative modeling of global context and local geometry is combined, depth ambiguity caused by insufficient geometric constraints under the condition of few samples is effectively solved, and the method is especially suitable for being used under the limited observation condition that only sparse source view angle images are provided. And performing high-quality new view rendering and image generation on an unknown scene.

In some implementations, the techniques described herein relate to a method including: identifying, by a processor, a digital image file that includes a background scene and an additional digital image file that includes a foreground object; compositing, by a machine learning model executed by the processor, the digital image file and the additional digital image file to produce a composite digital image file that includes the foreground object placed in front of the background scene by: identifying a location within the background scene in the digital image file for placement of the foreground object from the additional digital image file; transforming at least one aspect of the foreground object to harmonize with the background scene; and creating a composite image file that includes the harmonized foreground object in the location within the background scene; causing display, by the processor, of the composite image file.

The invention discloses a key frame screening and video question-answering method and device based on target existence and a storage medium, and the method comprises the steps: (1) carrying out the uniform sampling of an input video stream at a fixed sampling rate, and generating a frame sequence set with a continuous time sequence; (2) generating a target existence table according to the user question and the input video stream, and prompting the large language model to perform frame screening according to the target existence table to obtain candidate frames; (3) splicing the candidate frames into a single composite image according to a time sequence, inputting the spliced image and a question into a large language model, and outputting a refined key frame sequence; and (4) splicing the refined key frame sequence into a single composite image according to the image synthesis method in the step (3), inputting the spliced image and a user question into a large language model, and outputting a JSON format structured response containing answer options, reasoning interpretation and confidence score. By utilizing the method, fine-grained dynamic actions can be preserved, the frame screening sensitivity is eliminated, and the visual perception is enhanced.

The invention discloses a scenic spot interaction system based on virtual reality, and relates to the technical field of electronic business card production, the system comprises an interaction terminal and a server; the interaction terminal is provided with an NFC read-write chip, and the NFC read-write chip is used for establishing wireless communication with the target terminal equipment and receiving an image synthesis instruction of the target terminal equipment based on the wireless communication; the server responds to the image synthesis instruction, obtains a first target image containing a landscape and multiple tourists and generates an obtaining instruction of a second target image; the NFC read-write chip responds to the acquisition instruction, and a second target image containing the target person is acquired from the target terminal equipment; the server synthesizes the target person in the second target image into the first target image to obtain a third target image, and generates an electronic business card according to the third target image; and the NFC read-write chip sends the electronic business card to the target terminal equipment, so that the target terminal equipment accesses the third target image through the electronic business card. The group photo efficiency and convenience of tourists are improved.

A method, apparatus, non-transitory computer readable medium, and system for image processing include obtaining a text prompt describing an image element, generating an image embedding based on the text prompt, where the image embedding represents visual features of the image element, and generating a synthetic image depicting the image element based on the image embedding.

The invention discloses a cable joint defect automatic identification method and system for cable construction, and relates to the technical field of installation, maintenance and detection of power cables, and the method comprises the steps: obtaining an internal parameter, an external parameter and a distortion coefficient of a camera; adjusting the position of a camera according to a calibration result, and carrying out multi-angle shooting on the cable joint; carrying out geometric correction and color correction on the shot image; fusing the plurality of images through an image synthesis technology; extracting a key area of the cable joint from the corrected image; and determining space coordinates. And performing defect identification and classification by adopting a deep learning model, and performing quantitative measurement on the defect size. According to the invention, common defects in the cable joint can be accurately identified, the problems of missing detection and false detection in traditional manual detection are avoided, the automatic detection capability in electric power facility construction and maintenance is improved, safety monitoring, fault prevention and intelligent maintenance of an electric power circuit are facilitated, and the detection efficiency is improved. And the construction quality of cable joints and the long-term reliability of electric power facilities are ensured.

The invention relates to a high-speed target speed estimation method based on minimum entropy coarse and fine combination fast search. The method comprises the following steps: carrying out bandwidth synthesis on a baseband signal of a linear frequency modulation stepping echo signal; performing coarse velocity search on each sub-pulse, and performing phase compensation and coherent accumulation on each sub-pulse at each candidate coarse velocity; the waveform entropy of each candidate coarse velocity is calculated, and the candidate coarse velocity corresponding to the minimum entropy value in the waveform entropies is used as the optimal coarse velocity; compensating the intra-pulse Doppler distance coupling time shift of the baseband signal; performing distance pulse compression on the signal after coarse speed compensation; and performing bandwidth synthesis on the sub-pulse signals after pulse compression, performing fine grid search, performing phase compensation and coherent accumulation on each sub-pulse by using a fine phase compensation function for each searched fine velocity, and finally performing search according to a waveform entropy minimization principle to obtain an optimal fine velocity. According to the method, high-precision speed measurement and high-resolution range profile synthesis are synchronously completed in a single-frame signal.

The present disclosure relates to an image processing method and apparatus, an electronic device, and a storage medium. The method comprises: acquiring an original image; acquiring a target mask corresponding to the original image, the target mask being used for defining an image erasure region of the original image, and in the original image, the image erasure region comprising a target object; generating a target image on the basis of the original image and the target mask, the target image comprising a target region, and the target region being a mapping region of the image erasure region in the target image, wherein the target region does not comprise the target object; and synthesizing the target image and the original image to obtain a processed image.

It is possible to achieve both a high-resolution real-space image used in, for example, a video see-through type of AR device or MR device, and reduced system load. By an image acquisition unit, a first image captured at a first angle of view and a second image captured at a second angle of view narrower than the first angle of view within the first angle of view are acquired. By an image generation unit, a display image is generated by synthesizing an image of a high-resolution region based on the second image and an image of a peripheral region around the high-resolution region based on the first image. For example, based on gaze information of a user, the movement of an imaging direction of an image capturing unit for obtaining the second image is controlled, and the movement of a position of the high-resolution region is also controlled.

A method, apparatus, non-transitory computer readable medium, and system for generating style-matched images include obtaining a content prompt and a style prompt. The content prompt includes an object and the style prompt includes a style element. Embodiments then encode the content prompt and the style prompt to obtain a content embedding and a style embedding, respectively. Subsequently, embodiments apply a content mask to the content embedding and a style mask to the style embedding to obtain a weighted content embedding and a weighted style embedding, respectively. Embodiments then generate, using an image generation model, a synthetic image based on the weighted content embedding and the weighted style embedding. The synthetic image depicts the object from the content prompt and the style element from the style prompt.

The invention discloses a data-driven adaptive marketing picture generation system and method, and aims to solve the technical problem that a static template cannot intelligently adapt to dynamic content in the prior art, and the system comprises a data collection module which is used for obtaining business data and channel screenshots; the data integration module is used for calling a visual identification model, intelligently extracting target visual elements from the channel screenshot by taking the business data as a context, and formatting text data; the intelligent generation module is used for loading the dynamic template, filling the content, automatically calculating and adjusting the layout position of each element according to the actual rendering size of each element after filling, and dynamically generating a background image layer which is completely matched with the total size of the final content; and the image synthesis module synthesizes the content and the background image layer into a final marketing picture. According to the method, through content-driven adaptive typesetting and AI intelligent extraction, the bottlenecks of format stiffness and content matching difficulty are solved, and the efficiency, accuracy and visual quality of marketing picture automatic generation are remarkably improved.

The invention provides a poster image generation method and device, equipment, a medium and a product, and relates to the technical fields of large models, generative forms, generative models, Transformers, computer graphics, artificial intelligence, image synthesis, cloud computing and the like. According to the specific implementation scheme, initial cue words are generated according to an obtained poster, and text elements included in a to-be-generated poster image are determined; generating a text element layer according to the text element and the text element position corresponding to the text element, and generating an initial prompt word and an image element position according to the poster to generate an image element layer; wherein there is no overlapping position between the text element position and the image element position; and carrying out superposition processing on the text element layer and the image element layer to obtain a poster image to be generated. According to the method, on the premise of ensuring reasonable image-text typesetting layout, the computing resource cost required by poster image generation is reduced.

The embodiment of the invention relates to the technical field of image processing, and discloses a QNX-based image processing method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring timestamp data of a hardware clock, and injecting the timestamp data into each frame of original image data based on a hardware counter to generate standard image data; configuring a shared buffer area, and mapping the standard image data to a user space at least based on the shared buffer area; monitoring the load condition of the image processor GPU in real time; and dynamically adjusting the panoramic image synthesis mode at least based on the standard image data of the user space and the load condition of the GPU. The problem that the panoramic image cannot be generated in time due to serious GPU load is at least solved, the vehicle use safety of a user is guaranteed, and the vehicle use experience of the user is improved.

The invention relates to the technical field of image processing, in particular to a glass punching hidden crack detection method and system adopting image processing. Comprising the following steps: acquiring point cloud data of target glass, calculating the curvature of each region, and generating a curvature distribution diagram; simulating a light propagation path based on the distribution map, obtaining an illumination intensity adjustment coefficient, optimizing an illumination condition through adjustable light source equipment, and collecting an image data set; gaussian filtering is adopted to carry out de-processing on the image data, and then binarization processing is carried out to obtain a subfissure binary image; calculating length and width indexes of the crack according to the hidden crack binary image, determining a hidden crack position set, and aligning the hidden crack position set with the curvature distribution map to generate addition data; and finally, fusing the hidden crack position with the curvature distribution diagram to generate a preliminary report diagram, and generating a final hidden crack identification result through image synthesis. According to the invention, the detection precision of the hidden crack is effectively improved, and the method is especially suitable for detecting the hidden crack defect generated in the laser drilling process of a photovoltaic panel.

A multi-camera imaging system includes a plurality of imaging units arranged side-by-side to capture images of a scene. A calibration module is configured to determine intrinsic and extrinsic parameters of camera modules of the imaging units and to establish a per-pixel mapping from image coordinates to a three-dimensional space. An image processing module selects, for individual of the plurality of imaging units, a field-of-view region within a captured image and blends the fields-of-view from adjacent imaging units together to form a composite image.

The invention discloses a geometric image synthesis method, a related device and a system, and relates to the technical field of image synthesis, and the geometric image synthesis method comprises the steps: obtaining a geometric image demand text described by a user in a natural language; generating an image synthesis code according to the geometric image demand text by using the code large language model; synthesizing a plurality of geometric images based on the image synthesis codes; verifying the plurality of geometric images in combination with the geometric image demand text, and determining the conformity of the current image composite code according to the verification result; and if the conformity is greater than or equal to a preset conformity threshold value, outputting a geometric image which is verified to be qualified, and / or outputting an image synthesis code. According to the geometric image synthesis method disclosed by the invention, full-process automation from geometric image demand understanding to geometric image synthesis is realized, a user only needs to describe the geometric image demand through a natural language and does not need to have professional programming ability and graphic knowledge, and the technical threshold is greatly reduced.

The invention relates to the technical field of video noise reduction, in particular to an ultra-high-definition video noise reduction method. The method comprises the following steps: acquiring video data and noise variance; obtaining a gray threshold and a gradient threshold based on the gray value and the gradient value of each frame of image pixel point of the video data, obtaining the region of each frame of image, and then determining a growth criterion through the gray difference and the gradient difference to block the image; determining a gray feature weight and a texture feature weight based on the difference between the gray value and the gray limit value and the difference between the gradient value and the gradient limit value; improving the original DCT coefficient based on the feature weight and the texture feature weight to construct an exclusive dictionary, and denoising, decoding and reconstructing the image based on the exclusive dictionary; and synthesizing each frame of reconstructed image into a complete video. According to the invention, optimal noise reduction of the video image frame is realized and detail information of the video is reserved to the greatest extent.

A vision-based tracking control method for a quadruped robot includes receiving unilateral images captured by multiple cameras pointing different directions and utilizes an image stitching algorithm to synthesize the unilateral images into a panoramic image; based on the panoramic image, determining first and second position coordinates of the moving target and environmental obstacles; and formulating or modifying a navigation path for tracking the moving target based on the first and second position coordinates. The method significantly improves the perception ability, navigation accuracy and robustness of the quadruped robot in dynamic environments, thereby achieving more intelligent and flexible control.