499 results about "Image synthesis" patented technology

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.

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 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.

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 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.

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 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.

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.

The invention discloses a three-dimensional Gaussian splash reconstruction method for an underwater scene, and belongs to the technical field of computer vision and three-dimensional reconstruction. Comprising the following steps: acquiring a monocular video frame sequence of a target underwater scene, a corresponding camera pose sequence, an initial sparse point cloud, an initial three-dimensional Gaussian point set and learnable physical parameters of an underwater imaging model; in the training process, performing weighted evaluation on a reconstruction error based on a multi-view consistency mechanism of opacity weighting, calculating an importance score of each Gaussian point, and performing densification operation on a three-dimensional Gaussian point set; adopting a staged freezing strategy to cooperatively optimize the three-dimensional Gaussian point set and underwater imaging model parameters; and performing rendering and underwater image synthesis on any new view angle camera pose based on the optimized three-dimensional Gaussian point set and underwater imaging model parameters, and outputting a new view angle synthesized image to represent a reconstruction result. According to the method, the geometric compactness, the visual fidelity and the physical interpretability of an underwater three-dimensional reconstruction result are improved.

A method, apparatus, non-transitory computer readable medium, and system for data generation include obtaining a text prompt describing an object within a scene and generating, using a text-to-mask generation model and based on the text prompt, a color map corresponding to the scene. The color map indicates a region corresponding to the object from the text prompt. An image segmentation mask is generated based on the color map. The image segmentation mask comprises a plurality of regions corresponding to a plurality of image elements in the scene including the region corresponding to the object from the text prompt.

The invention relates to the technical field of image reconstruction, in particular to an intelligent trolley image compressed sensing reconstruction method and system based on deformable convolution. The method comprises the following steps: an intelligent trolley obtains multiple frames of original images in a driving scene through a vehicle-mounted image acquisition module and carries out preprocessing to generate a preprocessed image set; performing feature extraction, feature matching, hypergraph transformation and image synthesis operation on the basis of the preprocessed image set, performing compressed sensing sampling processing at the same time to obtain compressed sensing sampling data, and inputting the compressed sensing sampling data into a preset compressed sensing reconstruction network; a deformable convolution module is embedded in the front end of the compressed sensing reconstruction network for feature extraction and image detail recovery calculation, and a preliminary reconstruction image is generated; and embedding a deformable deconvolution module at the rear end of the compressed sensing reconstruction network to carry out pixel-level adjustment and quality evaluation until a final reconstruction image meeting the reconstruction effect requirement is generated. According to the invention, the precision of vehicle-mounted image compressed sensing reconstruction can be improved.

A computer-implemented is disclosed. The method includes: obtaining a composite image depicting an identifiable foreground object; performing image segmentation to isolate a foreground object from the composite image, the image segmentation yielding a segmented composite image; determining object attributes of the foreground object based on image analysis of the segmented composite image; obtaining at least one prompt for a text-to-image model based on the object attributes of the foreground object, the at least one prompt defining an intended replacement background for the composite image; providing, to the text-to-image model, instructions to generate a replacement background image for the composite image based on the at least one prompt; receiving, from the text-to-image model, a replacement background image for the composite image; and generating a new composite image, the generating including compositing portions of the composite image corresponding to the foreground object with the replacement background image.

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.

A method, apparatus, non-transitory computer readable medium, and system for modulating the level of fidelity to an input image include obtaining an input image and a fidelity parameter. The input image depicts an entity, and the fidelity parameter indicates a level of fidelity, i.e., faithfulness, to the input image. Embodiments then add noise to the input image based on the fidelity parameter to obtain an intermediate noise image. Subsequently, embodiments generate a synthetic image based on the intermediate noise image using an image generation model. The synthetic image includes a vectorizable depiction of the entity and has the level of fidelity to the input image indicated by the fidelity parameter. The vectorizable depiction is more suitable for conversion to vector format, as the resulting vector image will have a reduced number of paths and shapes.

The embodiment of the invention provides a fundus image synthesis method and device, electronic equipment and a storage medium, and belongs to the technical field of image processing. The method comprises the steps of obtaining a fundus center through target detection based on an original fundus image, and obtaining an initial light spot through light spot generation; configuring the light spot grade of the initial light spot based on the distance between the initial light spot and the fundus center; modifying light spot parameters of the initial light spot according to the light spot grade to obtain a target light spot; performing image synthesis on the target light spot and the original fundus image to obtain an intermediate fundus image; and performing image reconstruction based on the middle fundus image to obtain a target fundus image. According to the embodiment of the invention, the image quality of the eye fundus image after formation can be improved.

Approaches are presented for training an inverse graphics network. An image synthesis network can generate training data for an inverse graphics network. In turn, the inverse graphics network can teach the synthesis network about the physical three-dimensional (3D) controls. Such an approach can provide for accurate 3D reconstruction of objects from 2D images using the trained inverse graphics network, while requiring little annotation of the provided training data. Such an approach can extract and disentangle 3D knowledge learned by generative models by utilizing differentiable renderers, enabling a disentangled generative model to function as a controllable 3D “neural renderer,” complementing traditional graphics renderers.

A system for generating contrast-enhanced magnetic resonance images of a subject includes an input configured to receive at least one non-contrast enhanced image of the subject, and a contrast-enhanced magnetic resonance (MR) image synthesis neural network coupled to the input and configured to generate a contrast-enhanced magnetic resonance image of the subject based on the at least one non-contrast enhanced image of the subject. The contrast-enhanced MR image synthesis neural network is trained using a set of training data comprising at least quantitative data.

Embodiments of the present application provide an image synthesis method, device, system, electronic device and storage medium, and relate to the technical field of image processing. The method comprises: a first device receiving a first image containing a target object sent by a second device; generating target position data for the target object based on the first image; wherein the target position data indicates the position of the edge pixel points of the first image region occupied by the target object in the first image; the first device sends the target position data to the second device; after receiving the target position data, the second device obtains the pixel values of each pixel point contained in the first image region based on the target position data, and obtains a synthesis image by combining the pixel values of each pixel point contained in the second image region. In this way, the amount of data required for transmission can be reduced, the bandwidth required for transmitting data can be reduced, and the efficiency of image synthesis can be improved.

The invention provides a cigarette packet graphic element self-adaptive combination method and system, and relates to the field of image synthesis, and the method comprises the steps: collecting historical data and market change data, and employing time sequence analysis to obtain time sequence difference characteristics; according to the time sequence difference characteristics, K clustering is adopted to obtain a time-sensitive element cluster; screening out clusters with response index values higher than a preset threshold value according to the element clusters to obtain an adaptive cluster list; calculating a distribution uniformity index, screening out subsets of which the uniformity index is lower than a preset threshold value, and obtaining element subsets with balanced distribution; calculating the combination compatibility among the elements, and reserving subsets with the compatibility higher than a preset threshold value to obtain an adjusted combination scheme; and testing the response speed according to the adjusted combination scheme, and screening out the combination scheme of which the response speed reaches a preset threshold value to obtain an optimized combination version. The method can dynamically sense the time difference and intelligently adjust the element combination according to the time difference.

A learning system for executing learning of a generator of a generative adversarial network (GAN) which allows a user to control a plurality of features relating to a generated image, the learning system comprising at least one processor configured to: acquire a plurality of portion codes respectively corresponding to the plurality of features based on a latent code for generating the generated image and a plurality of mapping networks respectively corresponding to the plurality of features; generate the generated image based on image synthesis networks configured to generate the generated image through use of the plurality of portion codes; and execute the learning of the generator including the plurality of mapping networks and the image synthesis networks based on the generated image and a trained discriminator of the GAN.

Embodiments of the present application provide a parking method, device, system, storage medium, vehicle and mobile terminal. The parking method comprises: acquiring real-time environment information of a vehicle; the real-time environment information comprises real-time environment images at multiple observation points of the vehicle; determining a real-time automatic driving state of the vehicle according to the real-time environment information; when it is determined that the real-time automatic driving state is an abnormal state, controlling the vehicle to stop moving, and sending abnormal reminding information to a mobile terminal, so that the mobile terminal displays the abnormal reminding information; when a viewing request of the mobile terminal for the abnormal reminding information is received, sending the multiple real-time environment images to the mobile terminal after being synthesized into a real-time panoramic image, so that the mobile terminal displays the real-time panoramic image. Embodiments of the present application optimize the hardware resources of the vehicle, effectively reduce the program lag probability in the automatic driving process of the vehicle, and improve the accuracy and safety of automatic parking.

Proposed are a High Dynamic Range (HDR) image acquisition apparatus and an HDR image acquisition method. The HDR image acquisition apparatus may include at least two image acquisition modules configured such that each of the at least two image acquisition modules includes an image sensor and has a target sensing light intensity range different from each other, the at least two image acquisition modules being configured to simultaneously photograph a target photographing region. Furthermore, the HDR image acquisition apparatus may include an HDR-type image synthesis part configured to generate an HDR image by synthesizing at least two images in an HDR synthesis manner, the at least two images being simultaneously acquired by the at least two image acquisition modules.

This invention discloses a method, system, device, and storage medium for generating camouflage images, belonging to the field of computer vision and image generation. It employs a hierarchical automatic annotation method based on a large visual language model to obtain camouflage image annotation data, guiding the process through a three-level annotation workflow: online batch processing, offline supplementation, and manual review, outputting structured results. Based on a diffusion model framework, it integrates two types of feature enhancement modules—texture prior and color adaptation—to construct a generative model. After training with annotation data, it inputs the camouflage image to be restored, constraining texture consistency and aligning global color consistency through the two modules respectively, thus completing the generation and restoration of the camouflage image. This method for generating and restoring camouflage targets, combining semantic pairing, texture prior, and color consistency, along with its accompanying dataset, can be widely applied to camouflage target image synthesis, camouflage target detection enhancement, augmented reality, and other technical directions.

This invention relates to the field of augmented reality technology and discloses an augmented reality image synthesis method based on the optical properties of pearlescent pigments. The method includes: acquiring spectral reflectance data of pearlescent pigments under different illumination and viewing angles; analyzing the spectral reflectance data to calculate color correlation metrics under multiple angle combinations, thereby determining the delay parameter for signal transmission between angles; using the delay parameter and the distribution characteristics of the spectral reflectance data within a preset time interval to derive a gloss fluctuation index of the pearlescent pigment within the time interval; combining the gloss fluctuation index and the attribute differences of the spectral reflectance data after frequency domain transformation to evaluate the synthesis interference probability at each frequency point and extract key interference frequencies; integrating the distribution patterns of all key interference frequencies to generate an image rendering error probability; and finally adjusting the texture rendering parameters of virtual objects in the augmented reality scene according to the image rendering error probability to achieve the generation and output of the synthesized image.

The invention discloses an improved tire embossed character detection method. The method comprises the following steps: S1, sequentially photographing the periphery of a tire through an area-array camera to obtain four tire images; s2, synthesizing the four tire images into one tire imprinting character image by using a synthesis algorithm, performing coarse positioning on the synthesized image by using a circle detection algorithm, and cutting the synthesized tire imprinting character image into two rectangular images by using a cutting algorithm; s3, performing data annotation on the two cut rectangular images by using an annotation tool, and then performing data enhancement to expand a data set; s4, putting the prepared data set into a YOLOv5-based network for model training; and S5, identifying tire image embossed characters by using the trained model, and displaying a result on a software page in real time and storing the result in a database. According to the invention, the accuracy rate of automatic identification of the tire embossed characters reaches 98%, the detection efficiency is improved, the personal error is reduced, and the production quality and the production stability of products on the tire error-proofing assembly line are ensured.

A method includes accessing first image data captured by a camera of an environmental scene in which a subject arm performs a task. A first image is obtained from the first image data. The first image includes a subject arm object representing the subject arm and a background representing the environmental scene. A set of robot arm parameters for the first image is determined at least in part based on the subject arm object. An image of a robot arm object is rendered based on the set of robot arm parameters and a robot arm model. The image of the robot arm object is composited with the first image to obtain a synthetic robot image, which may be used for generating training data for AI model training.

Provided is a display image generation apparatus including a captured image acquisition section that acquires data of an image captured by a camera, an object arrangement section that arranges a virtual object to be operated by a user in a virtual three-dimensional space, a display image generation section that generates a display image by drawing an image of the virtual object and synthesizing the image of the virtual object with the captured image, and an output section that outputs data of the display image. The display image generation section switches whether or not to use an intermediate image representing the image of the virtual object from a viewpoint of the camera when drawing the image of the virtual object, according to a state of a display world including the virtual object.