940 results about "Intermediate image" patented technology

The invention provides a cross-domain target detection method and system based on semi-supervised learning in extreme weather, and the method comprises the steps: enabling a student model to learn bounding box regression knowledge of a trained teacher model through distillation, so as to achieve the regression capability in extreme weather; performing instance-level feature adversarial training on a pre-training detection model which is obtained in a distillation stage and has a regression capability in extreme weather by using an intermediate-state image which is generated by a diffusion model and is closer to a target domain and has smaller noise, and determining a new teacher model and a new student model based on the detection model after the adversarial training; in the fine tuning stage, a dynamic category pseudo label screening strategy is utilized to screen pseudo labels generated by a new teacher model, and a screened target domain pseudo label is utilized to train a new student model to obtain a target detection model used for detecting a to-be-detected image in extreme weather. The phenomena of leak detection and error detection in extreme weather scenes in the prior art can be reduced, and the original reasoning speed is kept.

The invention provides a diffusion model LoRA fine tuning optimization method and system based on CLIP loss and perception loss, and the method comprises the steps: 1, dynamically adjusting the weights of the CLIP loss and the perception loss in combination with the CLIP loss and the perception loss in a LoRA fine tuning process; 2, calculating the semantic similarity between the denoised intermediate image and the target text by using a CLIP model, and optimizing the noise prediction capability of a diffusion model according to the similarity difference; step 3, calculating the difference between the intermediate image and the target image in the feature space by using perception loss, optimizing the noise prediction capability of the diffusion model, and improving the visual quality and detail fidelity of the generated image; and step 4, adjusting whether to start CLIP loss and perception loss or not according to the training progress. By introducing the CLIP loss, the model can better align the image and the text in the fine tuning training process, so that the generated image better conforms to the description of the text prompt.

The invention discloses a wafer detection scattered light collection objective lens, and relates to the technical field of optical imaging. The scattered light collecting lens is applied to wafer defect detection, the scattered light collecting lens sequentially comprises a first lens group, a second lens group, a third lens group and a fourth lens group from an object plane to an image plane in the optical axis direction of the lens, twelve lenses and an aperture diaphragm, and the twelve lenses and the aperture diaphragm are coaxially arranged; the first lens group is a catadioptric lens group, has positive focal power, and is used for imaging an object plane to an intermediate image plane through two times of reflection. The numerical aperture of the acquisition objective lens provided by the invention is equal to 0.9, the field of view is equal to 2.6 mm, the acquisition objective lens has the characteristics of large field of view, large numerical aperture and small center shielding, the capability of detecting scattered light of defects can be greatly improved, chromatic aberration is corrected by adopting a refraction and reflection mode, and a wider laser bandwidth can be realized compared with a general refraction mode of combining CaF2 and fused quartz.

A method includes determining at an extended reality (XR) device, a first set of vertex adjustment values of a distortion mesh and receiving image frame data of a scene captured at a first time and at a first head pose using a see-through camera of the XR device. The method further includes applying the first set of vertex adjustment values of the distortion mesh to the image frame data to obtain intermediate image data, and predicting a second head pose at a second time subsequent to the first time. The method also includes generating, based on the predicted second head pose, a second set of vertex adjustment values of the distortion mesh, applying the second set of vertex adjustment values of the distortion mesh to the intermediate image data to generate a rendered virtual frame and displaying the rendered virtual frame by the XR device at the second time.

The invention discloses a low-light image enhancement method based on low-rank recovery and deep diffusion fusion, and relates to the technical field of image processing. Comprising the following steps: inputting a to-be-processed low-light image into a trained low-light image enhancement model, and transforming a pixel value of the to-be-processed low-light image through a discrete cosine transform module to obtain a spectrogram in a spectral domain; performing forward noise addition at each time step through a Markov chain mechanism, capturing global features of noise distribution through a Transform network, and removing noise components in the current time step according to the global features of the noise distribution to obtain an intermediate image; taking the intermediate image corresponding to the last time step as a new spectrogram; and performing inverse discrete cosine transform on the new spectrogram through an inverse discrete cosine transform module, and converting the new spectrogram from the frequency domain to the spatial domain to obtain an enhanced low-light image. The method pays attention to the globality of noise components, and can improve the definition and quality of the enhanced low-light image.

The invention provides a digital human image generation method, device and equipment and a readable storage medium, and the method comprises the steps: generating a plurality of frames of intermediate images matched with driving voice data through model driving, and combining a pre-trained tooth generation network with the first K frames of correction images for completing tooth reconstruction of the intermediate images, according to the invention, tooth reconstruction is carried out on the current frame of intermediate image, the tooth area on the intermediate image is masked, and the corrected image corresponding to the intermediate image is fused with the intermediate image according to the masked image, so that the consistency matching of the facial expression on the intermediate image and the lip mouth shape with the driving voice data is ensured, and the user experience is improved. According to the method, the optimization of the tooth area on the intermediate image is realized, so that the teeth of the digital human are more clear and vivid visually and present a natural and reasonable arrangement state, and the overall sense of reality of the digital human is improved.

A method, apparatus, non-transitory computer readable medium, and system for video generation include obtaining an input image having an element depicted in a first view angle, generating a synthetic image depicting the element of the input image from a second view angle different from the first view angle, generating an intermediate image by interpolating based on the synthetic image, and generating a video based on the synthetic image and the intermediate image, where the video depicts the element of the input image from a changing view angle.

A plenoptic imaging device for forming an image of the light field of an object, includes a sample holder suitable for holding the object; a main optical assembly, arranged to focus rays of the light field of the object, at a distance z0, referred to as the object distance, from the sample holder, which distance is smaller than an object focal distance f1 from the main optical assembly so as to form an intermediate optical image (Iv) of the object in an intermediate image plane at a distance z1 from the optical assembly; an assembly for sampling the light field, which assembly is suitable for acquiring spatiodirectional information on the rays forming the virtual image and for forming the image of the light field.

Described herein are systems, methods, and instrumentalities associated with using a multi-layer perceptron (MLP) neural network to process medical images of an anatomical structure. The processing may include padding an input image in accordance with the training of the MLP neural network, splitting the input image (e.g., the padded input image) into patches of a same size, and processing the patches through the MLP neural network over one or more iterations. During an iteration of the processing, the patches may be processed separately and re-combined into an intermediate image before the intermediate image is shifted to concatenate portions of the image that are derived from different patches. This way, global features of the anatomical structure may be learned and used to improve the quality of the image generated by the MLP neural network, without incurring significant computation or memory costs.

A display device includes a driving frequency calculator configured to divide output image data into central, intermediate, and peripheral images respectively corresponding to central, intermediate, and peripheral visual field regions, calculate spatial frequencies of the central image, the intermediate image, and the peripheral image, calculate critical flicker frequencies of the central visual field region, the intermediate visual field region, and the peripheral visual field region based on the spatial frequencies and based on viewing angles of the central visual field region, the intermediate visual field region, and the peripheral visual field region, and determine driving frequencies of the central visual field region, the intermediate visual field region, and the peripheral visual field region based on the critical flicker frequencies, and a display panel configured to display an image based on the output image data and based on the driving frequencies.

The invention discloses a four-channel fluorescence detection light path module of a super-resolution biological microscope, which divides a fluorescence signal of an intermediate image plane (5) into four conjugate channels (7a-7d) through multi-stage dichroscope light splitting, reflector guiding and lens focal plane adjustment on the premise of not introducing additional aberration, and realizes real-time synchronous detection by using a single large-target-surface camera (8). By adjusting the distance between the reflectors, the inclination angle of the parallel plate glass or the position of the lens, the device can adapt to different camera target surface sizes and compensate chromatic aberration, and the problems of motor switching delay and dual-channel aberration in the prior art are solved.

The invention belongs to the technical field of image processing, and particularly relates to a bridge pier intelligent inspection method, which comprises the following steps of: performing Gaussian filtering on an inspection image through Gaussian filters with different scales, and subtracting an original inspection image from obtained background images with different scales to obtain intermediate images with different scales; calculating the enhancement degree of the gradient features of the pixel points in the intermediate images under different scales compared with the gradient features of the pixel points in the original inspection image; calculating the retention degree of the multiple LBP features of the pixel points in the intermediate image under different scales compared with the original inspection image; and according to the enhancement degree and the retention degree, calculating the weights of the gray values of the pixel points in the intermediate image under different scales, and carrying out weighted summation on the gray values of the pixel points in the intermediate image under different scales so as to obtain a uniform light image. According to the invention, the finally obtained uniform light image achieves better balance in detail and texture aspects.

The invention provides an image protection method and device based on a diffusion model, equipment and a medium, and relates to the technical field of face privacy security protection, and the method comprises the steps: carrying out the feature extraction of an obtained target image through employing a BLIP-2 model, obtaining a target image feature vector, a text description, and a text feature vector corresponding to the text description, the method comprises the following steps: obtaining a target image feature vector, taking text description as a private key prompt word Prompt1, inputting the target image feature vector and the text feature vector into a target comparison learning model, updating the private key prompt word Prompt1, obtaining an intermediate image output by a sending end, and transmitting the intermediate image and the updated private key prompt word Prompt1 to a receiving end. And the controllability of the target image in the encryption process and the security of the target image in the transmission process are improved.

Disclosed in the present invention are an image stitching method and apparatus, an image processing method, and a readable storage medium. The image stitching method comprises: according to a preset arrangement sequence, performing row arrangement and column arrangement of a plurality of unit images to be stitched; determining one of a row arrangement direction and a column arrangement direction as a first direction, and the other one as a second direction; determining seams between adjacent images among the plurality of unit images in the first direction, and stitching adjacent unit images according to the seams, so as to obtain a plurality of intermediate images; and determining seams between adjacent images among the plurality of intermediate images in the second direction, and stitching adjacent intermediate images according to the seams, so as to obtain a result image. In the present invention, images are stitched in two-dimensional directions by means of generating a seam network, thereby avoiding the problems such as artifacts, misalignments, overlapping and blank spaces that are generated during image stitching.

A system and a method for PET imaging may be provided. Scan data of an object collected by a PET scan over a scan time period may be obtained. A plurality of target sets of scan data may be determined from the scan data based on a preset condition. Each of the plurality of target sets of scan data may correspond to a target sub-time period in the scan time period. One or more intermediate images corresponding to one or more sub-time periods different from the plurality of target sub-time periods in the scan time period may be generated based on the plurality of target sets of scan data. A target image sequence of the object may be generated based on the plurality of target sets of scan data and the one or more intermediate images.

Disclosed in the embodiments of the present disclosure are a video generation method, an apparatus, an electronic device and a storage medium. The method comprises: during an image editing process of subjecting an initial image to a series of image editing operations to obtain a target image, acquiring initial key frames, the initial key frames being intermediate images between the initial image and the target image during the image editing process, the different initial key frames being intermediate images generated when different image editing operations of the series of image editing operations are triggered during the image editing process, and the intermediate images showing editing effects corresponding to the image editing operations; and generating a target video according to the initial key frames so as to record the image editing process of the target image. The present disclosure can be used for recording an image editing process.

A multimodal microscope having a light source for transmitting excitation light, an illumination beam path having a microscope objective for guiding the excitation light onto and / or into a sample, at least one first spatial light modulator being present in the illumination beam path for manipulating the excitation light, a detector for detecting detection light emitted by the sample owing to illumination with the excitation light, a detection beam path, comprising the microscope objective or a further microscope objective, for guiding the detection light onto the detector, and a controller at least for controlling the first spatial light modulator. The multimodal microscope includes a switchable optical functional group, which is switchable at least into a first and a second switching state, and the first spatial light modulator is situated either in or in the vicinity of an intermediate image plane or a pupil plane of the illumination beam path, depending on the switching state.

The invention discloses a machine vision-based automatic identification and calculation method for pore change of a rock section under an SEM (scanning electron microscope), and the method comprises the steps: obtaining original image data of the rock section through a scanning electron microscope, transmitting the original image data to a machine vision SEM pore intelligent quantification cloud platform, and dividing the original image data into a plurality of image sub-regions which are not overlapped with each other; calling a special threshold inversion model to analyze the gray information of each sub-region, and determining a dual-phase pore segmentation threshold; carrying out edge extraction on the sub-regions after threshold processing by adopting a pore edge gradient enhancement algorithm; performing topological structure analysis on the intermediate image data by using a biphase pore topological inversion model, constructing and optimizing a pore space topological network, and eliminating false pore structure data; and according to the optimized pore space topology network, integrating calculation results of all the sub-regions to obtain a pore change identification calculation result of the whole rock section. The method improves the pore identification precision and processing efficiency, and adapts to different types of rock samples.

The subject technology receives an input image, the input image comprising a selfie. The subject technology transforms, using a neural network, the input image to a latent representation of an identity. The subject technology transforms, using a diffusion model, a text condition to a second latent representation compatible with the latent representation of the identity. The subject technology transforms a pose template to a set of latent features for the diffusion model. The subject technology generates an intermediate image based on the latent representation of the identity, the second latent representation, and the set of latent features. The subject technology modifies, using a face enhancement network, the intermediate image based on the input image. The subject technology generates, using a face restoration network, a final output image based on the modified intermediate image. The subject technology provides for display the final output image on a display of a client device.

A method for digitizing image-based data includes receiving an image file including one or more target objects, generating an intermediate image by removing noise from the image file using a denoising machine learning model, identifying the one or more target objects included in the intermediate image using an object segmentation machine learning model, discretizing the one or more target objects that were identified using the trained object segmentation machine learning model, and storing the one or more target objects that were discretized in a data file, visualizing the one or more target objects, or both.

The invention provides an image processing method and device, computer equipment and a storage medium. The method comprises the steps of simulating an initial convolution operation executed by a super-resolution model on an original image based on a plurality of first lookup tables LUT to obtain an intermediate image; executing an operation of the super-resolution model for the intermediate image to obtain a super-resolution image; wherein the initial convolution operation comprises a rising channel convolution operation and a point convolution operation which are sequentially and linearly executed, the convolution kernel size of the rising channel convolution operation is larger than that of the point convolution operation, and the positions of the convolution kernels of the point convolution operation and the rising channel convolution operation are in one-to-one correspondence; the plurality of first LUTs comprise one-dimensional LUTs in one-to-one correspondence with positions of convolution kernels of the ascending channel convolution operation, the one-dimensional LUTs are used for recording a mapping relation between values of pixels and point convolution results, and the point convolution results are products of the values of the pixels, weights of corresponding positions in the convolution kernels of the ascending channel convolution operation and weights of the convolution kernels of the point convolution operation.

The invention discloses a lightweight high-speed photographing system and method based on an event camera, and the system comprises a hybrid imaging module which is used for synchronously collecting the image data and event data of a scene, and comprises a visible light camera and an event camera; the hardware synchronization module is electrically connected with the visible light camera and the event camera, and is used for receiving a frame synchronization signal of the visible light camera and sending a trigger signal to the event camera, so that the event camera inserts a time stamp corresponding to a visible light image frame exposure moment in an event stream; the data processing and fusion unit is in communication connection with the hybrid imaging module and is used for receiving the image data and the event data and executing the following steps: performing time alignment on an event stream and a visible light image frame stream based on a time stamp; operating a high-frequency frame insertion reconstruction algorithm, and reconstructing a multi-frame intermediate image by using event stream data between two frames; synthesizing the original visible light image frame and the reconstructed intermediate image into a high-frame-rate video stream; and the power supply and interface module is used for supplying power to the system and outputting a high-frame-rate video stream.

2.1. The simultaneous acquisition of multiple images in the microscope should be made possible with greater flexibility. Furthermore, simultaneous acquisition should be made possible with less effort. 2.2. In a microscope (1) with a main beam splitter (22) and an adjustable beam deflection unit (21) for moving the illuminating light through the sample space (P), the beam deflection unit (21) is arranged optically between the light source (12) and the main beam splitter (22) for this purpose, so that the sample light reaches the sensor (28) away from the beam deflection unit (21), wherein both the illuminating light and the sample light pass through the same intermediate image (ZB).

The invention discloses a position identification method of an ultrasonic puncture needle and ultrasonic equipment, and belongs to the technical field of ultrasonic image processing, the method comprises the following steps: in a current detection period, collecting multiple frames of ultrasonic images from a video of real-time ultrasonic puncture operation, and when each frame of ultrasonic image is collected, determining the position of the ultrasonic puncture needle; determining a first image based on the currently acquired ultrasonic image and a target ultrasonic image corresponding to the currently acquired ultrasonic image, wherein the target ultrasonic image is the previously acquired ultrasonic image; the first image is used for representing the motion displacement between the currently collected ultrasonic image and the target ultrasonic image, generating a second image based on the gray value of each pixel point in the first image and the gray value of the corresponding pixel point in the intermediate image, and determining the target ultrasonic image based on the gray value of each pixel point in the second image and the needle inserting direction of the puncture needle. And determining the needle body area of the puncture needle in the second image, so that the puncture needle can be detected in real time aiming at the ultrasonic image, and the needle body position of the ultrasonic puncture needle can be positioned more accurately.

The invention discloses an optical quasi-three-dimensional space converter which can convert a three-dimensional space into a large number of fixed-position two-dimensional real images with different object image distance related parameters. The optical quasi-three-dimensional space converter comprises a zoom long-focus lens, a short-focus micro lens array and an object image distance related parameter regulator; one of the two lenses is used for receiving light from an object space and imaging the light into an intermediate image; one is used for converting the intermediate image into a virtual image or a real image of a three-dimensional space; the intermediate image horizontal turner turns the intermediate image left and right in the horizontal direction for real image space conversion; and the object image distance related parameter regulator is used for regulating the focal lengths of the zoom long-focus lens, the short-focus micro lens array and the pick-up lens and the relative positions of the lenses. The optical quasi-three-dimensional space converter can be used for a wide-field-of-view three-dimensional camera, a wide-field-of-view three-dimensional projector for naked eye observation, a three-dimensional optical processing machine, artificial eyes and the like.

The invention relates to the technical field of operation state detection of traction substation equipment, and discloses an intelligent image recognition method and system for the operation state of the traction substation equipment and a medium, and the method comprises the steps: obtaining initial image data of the operation state of the substation equipment; preprocessing the initial image data to obtain intermediate image data; obtaining a sample actual label from the intermediate image data, and taking the intermediate image data and the corresponding sample actual label as a group of training samples so as to form a training sample set; constructing a state recognition model based on a deep learning neural network, optimizing weights and thresholds of a feature extraction unit and a recognition unit based on a particle swarm algorithm in combination with a training sample set, and finally completing training of the state recognition model; and detecting the intermediate image data of the operation state of the substation equipment through the trained state recognition model. Through the method provided by the invention, the workload of operation and maintenance personnel is reduced, the inspection efficiency is improved, and the risk occurrence probability is reduced.

The invention relates to a cross-modal retrieval technology, in particular to a cross-modal image-text pedestrian retrieval method based on a generative model, which comprises the following steps: acquiring text description information and a corresponding original pedestrian image; calling a diffusion generation model to generate an intermediate image based on the text description; original image features and text features are extracted through an image encoder and a text encoder respectively, and semantic-fused text feature representation is constructed; calculating the similarity between the image feature representation and the text feature representation to obtain an image-text matching score; introducing and generating an intermediate image as a semantic bridge, and realizing multi-modal fusion among the image, the text and the intermediate image through a cross attention mechanism to obtain fused image and text features; and using the fusion features to train a recognition model, and finally outputting a pedestrian image retrieval result corresponding to the text description. According to the method, the image-text alignment precision can be improved, the robustness of a scene with incomplete text description is enhanced, and the recognition accuracy in a cross-modal retrieval task is improved.

Disclosed in the present disclosure are an image processing method and apparatus, and a computer device. The method comprises: acquiring an image set; clustering images in the image set to obtain multiple initial image clusters, wherein the multiple initial image clusters are respectively configured to represent sets of images of different objects, and the same initial image cluster represents a set of images of the same object; determining at least two intermediate image clusters from among the multiple initial image clusters; and fusing the at least two intermediate image clusters to obtain an image clustering result of the image set.

The invention provides an image enhancement method, system and device for an automatic driving vision task, and belongs to the technical field of combination of computer vision and automatic driving, and the method comprises the steps: obtaining a multi-view image of a to-be-detected target, and splicing adjacent view images; applying a first size grid mask on the spliced image; removing a mask in an overlapping region of the spliced image to obtain an intermediate image; detecting a preset target area in the intermediate image, and applying a second size grid mask to the preset target area; and replacing the first size grid mask at the corresponding position with the second size grid mask to obtain an enhanced image. Based on the method, the invention further provides an image enhancement system for the automatic driving vision task. According to the method, a multi-view grid mask enhancement method is adopted, the first-size mask and the second-size mask are used for carrying out global subject covering and local small target covering respectively, and the robustness and generalization ability of a visual perception model in an automatic driving system in a complex environment are improved.

Disclosed are systems, apparatuses, processes, and computer-readable media for processing one or more images. For example, a method includes obtaining an inpainted image based on providing a first image to an ML model; combining the first image and a first auxiliary image of the first image into an intermediate image; obtaining an inpainted intermediate image based on providing the intermediate image to the ML model; and generating a second auxiliary image from the inpainted image and the inpainted intermediate image.