480 results about "Image transformation" patented technology

The invention discloses a generative invisible watermarking method based on a stable diffusion model, which comprises the following steps of: processing a text prompt / image prompt to obtain a latent variable, guiding to generate an image by using the latent variable, converting the image into a low-dimensional potential representation through a variational auto-encoder, generating potential noise according to the low-dimensional potential representation, and generating a stable diffusion model according to the potential noise. Performing multiple iterations by combining the potential noise and the low-dimensional potential representation to generate a noise-containing image; a watermark decoder is utilized to finely adjust the decoder, so that the finely-adjusted weight watermark is embedded into the generated image; and meanwhile, when the stable diffusion model generates an image, compressing the image into potential noise by using a variational auto-encoder, then performing wavelet transform on the obtained potential noise to obtain a noise watermark, mapping noise watermark information into an annular pattern watermark, embedding the annular pattern watermark into a noise low-frequency sub-band, and performing image generation by using the watermark-containing noise to obtain a watermark-containing image. According to the method, the watermark is embedded in the image generation process, and the watermark information in the generated image can be extracted.

Denoised magnetic resonance images are generated using a two-step process. An initial set of images is first denoised on a per channel basis using a locally low-rank-based denoising technique. The denoised coil channel images are transformed back into k-space and the denoised k-space data are then applied to a nonlinear image reconstruction. In some instances, the nonlinear image reconstruction can be implemented using a trained neural network. The neural network may be trained using a self-supervised learning technique.

The invention discloses a cross-modal pedestrian re-identification method based on a multi-scale cross attention Transform, and the method comprises the following steps: step 1, image acquisition, step 2, feature set acquisition, step 3, feature fusion, and step 4, feature determination: in an initial stage of a network, two standard data sets are used to acquire RGB and infrared images, and then, the RGB and infrared images are subjected to feature fusion; the method comprises the following steps: extracting key features from RGB and an infrared image by using MSCAViT to obtain two feature sets, then converting the RGB image into a grayscale image, inputting the grayscale image into an MSCAViT model to obtain a third feature set, determining an optimal weight by using an OSCSO method through weighted fusion of the three feature sets, finally obtaining a fused feature, inputting a result feature into a decoding end of the MSCAViT, and finally obtaining a final image through the decoding end of the MSCAViT. According to the method, features finally used for image matching are generated, the model is evaluated through multiple indexes, compared with a traditional method, and the result shows that the proposed method shows a more excellent effect on solving the modal difference problem in a VI Re-ID task.

The invention discloses an image enhancement method based on multi-modal enhanced grid masking and fast Fourier convolution, and the method comprises the steps: S1, collecting the foresight sonar data of an underwater typical target, and making an initial data set; s2, performing data enhancement in a multi-mode shielding mode on the original foresight sonar image; s3, performing data enhancement of a traditional image transformation mode on the multi-mode shielding frame enhanced image, simulating changes of illumination, noise, visual angles and the like in different environments, and generating diversified training data; s4, constructing an image enhancement network based on fast Fourier convolution, and improving the processing and enhancement capability of the occlusion region by using a global receptive field; and S5, enhancing the image by using the trained model to obtain a clear target occlusion image with rich details. According to the method, through the data enhancement method in combination with the multi-mode shielding mode and the network based on the fast Fourier convolution, the image enhancement quality and efficiency of the severely shielded target of the forward-looking sonar are remarkably improved, and the method has good generalization ability and practical value.

The invention discloses a thyroid tumor diagnosis method and system based on ultrasonic and cytological image conjoint analysis. The method comprises the following steps: converting a thyroid B ultrasonic image of a patient to be diagnosed into an image feature vector FUS; performing structured feature extraction on the thyroid cytological image of the patient to be diagnosed to obtain a structured feature vector FCYTO; the image feature vector FUS and the structured feature vector FCYTO are converted into a fusion Token sequence; and inputting the Token into a multi-modal feature fusion and prediction network based on a Transform architecture, and carrying out feature fusion and classification prediction so as to obtain the probability that the thyroid tumor of the patient to be diagnosed is malignant. The thyroid tumor diagnosis based on multi-modal fusion is carried out on the basis of ultrasonic and cytological images, so that the diagnosis accuracy is improved.

Described are systems and methods for training a machine-learning model to generate image of biological samples, and systems and methods for generating enhanced images of biological samples. The method for training a machine-learning model to generate images of biological samples may include obtaining a plurality of training images comprising a training image of a first type, and a training image of a second type. The method may also include generating, based on the training image of the first type, a plurality of wavelet coefficients using the machine-learning model; generating, based on the plurality of wavelet coefficients, a synthetic image of the second type; comparing the synthetic image of the second type with the training image of the second type; and updating the machine-learning model based on the comparison.

The invention discloses a crack image processing method based on a neural network and plane geometry image transformation, and particularly relates to the field of general image data processing driven by plane geometry image transformation and neural network identification. Comprising the steps that an image collection platform projects a structured image coding pattern to the surface of a to-be-measured target through a preset image coding pattern template, and synchronously collects an image frame containing an image mark so as to construct a space anchor point used for image geometric reference in an image plane; view angle parameter change information in an image acquisition process is acquired through an image acquisition platform, and projection configuration parameters of an image coding pattern are combined. By constructing an image offset modeling mechanism jointly driven by an image coding pattern and a space anchor point, image in-plane correction of geometric distortion features in an image frame is realized, and structural consistency and texture continuity before subsequent image analysis are improved.

The invention provides a pipeline defect assessment method and system based on multi-source data fusion, and the method comprises the following steps: collecting the data of the operation state of a pipeline body through a related collection device, and marking the state and fault type of the pipeline, so as to construct a training set and a test set; and preprocessing the static data of the pipeline, wherein the preprocessing comprises data cleaning, standardization and feature extraction. And optimizing the extreme learning machine by using a bird flock algorithm, analyzing static data and evaluating pipeline defects. Preprocessing the image data of the pipeline body, wherein the preprocessing comprises image conversion, enhancement and denoising; and evaluating the circumferential weld defect based on the pre-trained VGG16 model. And performing decision-level fusion on a defect evaluation conclusion obtained by static data analysis and circumferential weld image analysis by using a D-S evidence reasoning algorithm to obtain a final evaluation result. The method effectively improves the accuracy and reliability of pipeline defect evaluation, and has wide application prospects and important practical value.

Systems and methods in accordance with embodiments of the invention are disclosed that use super-resolution (SR) processes to use information from a plurality of low resolution (LR) images captured by an array camera to produce a synthesized higher resolution image. One embodiment includes obtaining input images using the plurality of imagers, using a microprocessor to determine an initial estimate of at least a portion of a high resolution image using a plurality of pixels from the input images, and using a microprocessor to determine a high resolution image that when mapped through the forward imaging transformation matches the input images to within at least one predetermined criterion using the initial estimate of at least a portion of the high resolution image. In addition, each forward imaging transformation corresponds to the manner in which each imager in the imaging array generate the input images, and the high resolution image synthesized by the microprocessor has a resolution that is greater than any of the input images.

Described are systems and methods for training a machine-learning model to generate image of biological samples, and systems and methods for generating enhanced images of biological samples. The method for training a machine-learning model to generate images of biological samples may include obtaining a plurality of training images comprising a training image of a first type, and a training image of a second type. The method may also include generating, based on the training image of the first type, a plurality of wavelet coefficients using the machine-learning model; generating, based on the plurality of wavelet coefficients, a synthetic image of the second type; comparing the synthetic image of the second type with the training image of the second type; and updating the machine-learning model based on the comparison.

The invention discloses a multi-scale linear array camera splicing method and system based on point cloud. The method comprises the following steps: completing acquisition and preprocessing of point cloud data and image data of a target area; determining an overlapping region range between adjacent images; extracting spatial structure characteristics in the point cloud data, and performing multi-scale hierarchical decomposition on the point cloud through a multi-scale segmentation method; meanwhile, multi-scale image feature extraction is carried out on the images of the linear array camera; solving gradients in X and Y directions by adopting an optical flow method aiming at any pixel in the overlapping region, and calculating a motion vector between the pixels; fusing the optical flow information obtained under each scale, and constructing a globally consistent optical flow vector field; according to the fused optical flow vector, calculating to obtain a geometric transformation matrix of the whole overlapping region; and after image transformation and alignment are completed through the transformation matrix, fusion processing is carried out on overlapped areas. And the unification of the visual effect and the spatial integrity of the spliced image is ensured.

The invention discloses a multi-modal image registration method based on self-modal correlation and cross-modal estimation. The method comprises the following steps: firstly, constructing an initial frame model composed of image frequency decomposition, self-modal similarity coding, cross-modal matching estimation, optical flow generation and refinement and an image transformation module, and then performing spatial transformation on registered visible light and thermal infrared images to construct a training set; and training the model to obtain a visible light and thermal infrared registration model, and performing multi-modal image registration on visible light and thermal infrared images through the visible light and thermal infrared registration model. According to the method, the self-modal similarity of the visible light and thermal infrared images can be fully constructed, the accuracy of cross-modal feature matching estimation is improved, and high-precision registration of the visible light and thermal infrared images with remarkable view angle difference and modal difference is realized; and in addition, the visible light and thermal infrared images can still be stably registered under severe environment conditions such as night, rainy days and foggy days, and the robustness is relatively high.

The invention discloses a multi-scale Retinex low-illumination image enhancement method based on modified bilateral filtering, and belongs to the field of digital image processing. Aiming at the problems of insufficient image brightness, detail loss, noise interference and color distortion in a low-illumination environment, the invention provides an improved image enhancement algorithm. First, the RGB image is converted into an HSV color space, and the luminance component (V) is separated for subsequent processing to avoid color distortion caused by direct processing of the RGB channel. And then, a corrected bilateral filtering function is adopted as a center surrounding function of a multi-scale Retinex method, a brightness component is enhanced, a halo phenomenon is suppressed, and meanwhile, detail enhancement of a dark region and excessive suppression of a bright region are realized in combination with an improved Gamma transformation function. And for the reflection component, non-local mean filtering denoising is utilized, and image edge details are enhanced in combination with a Laplace operator. In addition, the color of the enhanced image is ensured to be natural and consistent by adaptively adjusting the saturation component. Experimental results show that the brightness and definition of low-illumination images can be remarkably improved, detail information and visual impressions are improved, and the method is suitable for various scenes such as night monitoring, medical image processing and agricultural monitoring. The method has the advantages of being excellent in enhancement effect, high in halo suppression and small in color distortion.

The invention discloses an image contour recognition method and device for a cell cluster, and relates to the technical field of biological image processing, and the method comprises the steps: converting an input cell cluster image into a binary image; carrying out expansion operation and distance transformation on the image after morphological operation, and carrying out subtraction operation on the two obtained images to obtain a third image containing an unknown region; detecting connected regions in the accurate foreground region, and marking each independent connected region, background region and unknown region to obtain a marked image; performing region segmentation on the original image to obtain segmented marked regions, generating a corresponding binary mask image, and extracting the contour of each marked region by using the binary mask image; identifying and filtering the contour of each marked area to obtain an effective contour; the size of each effective profile is converted into an equivalent sphere diameter. According to the invention, accurate segmentation of adjacent or adhesion areas in the image can be efficiently realized, and accurate counting and size measurement of cell clusters can be obtained.

The invention belongs to the technical field of computer vision and image processing, and particularly discloses a video defogging method and system based on enhanced alignment, and the method comprises the steps: carrying out the color mapping of a plurality of foggy images through a color lookup table, and obtaining a plurality of enhanced images; based on the enhanced image corresponding to the foggy target frame and the enhanced image corresponding to each foggy reference frame, performing space-time optical flow estimation to obtain a three-dimensional optical flow; performing image transformation on the foggy reference frame based on the three-dimensional optical flow to obtain an alignment frame; and based on the alignment frame, the foggy target frame and the enhanced image corresponding to the foggy target frame, performing splicing and feature fusion to obtain a defogged image corresponding to the foggy target frame. According to the method, details of the foggy image are enhanced, a more accurate alignment result is obtained, more inter-frame information is provided through accurate alignment frames, the network can be effectively helped to learn a better defogging capability, the utilization capability of the deep neural network on the inter-frame information can be improved, and a better video defogging effect is achieved.

The invention provides an image defogging enhancement and intelligent perception collaborative optimization method based on stream matching, and the method comprises the steps: obtaining a foggy input image, and constructing an ordinary differential equation for defining image transformation; the input image is input into a fog perception vector field, the fog perception vector field comprises an atmospheric scattering purifier and a defogging perception color lookup table, and the atmospheric scattering purifier is used for extracting multi-scale features from the foggy input image to generate clearer output; the defogging perception color lookup table adaptively adjusts the image color through a nonlinear color transformation mechanism; and solving the ordinary differential equation through an RK4 solver, iteratively updating the foggy input image at each time step, and ensuring that the input image can be stably converted into a clear image from an initial foggy state. The technical problems that in the prior art, the existing method is forced to reduce the model capacity due to the calculation efficiency constraint in the ultra-high-definition image defogging, so that the fog concentration estimation is incomplete, and the perception quality and the calculation efficiency cannot be considered at the same time are solved.

A system and method for end-to-end semi-supervised object detection is provided. The system retrieves labeled and unlabeled images from an image dataset and generates an input batch by application of image transformation(s) on the images. The system further generates a first result for each image of the input batch by application of a teacher neural network on the input batch. For an object in an unlabeled image of the batch, the first result includes candidate bounding boxes and scores for the boxes. The system determines a threshold score based on the scores and selects a foreground bounding box from the candidates. The system generates a second result by application of a student neural network on the unlabeled image and computes a training loss over the input batch based on the foreground bounding box and the second result. The system trains the student neural network based on the training loss.

Described are systems and methods for training a machine-learning model to generate image of biological samples, and systems and methods for generating enhanced images of biological samples. The method for training a machine-learning model to generate images of biological samples may include obtaining a plurality of training images comprising a training image of a first type, and a training image of a second type. The method may also include generating, based on the training image of the first type, a plurality of wavelet coefficients using the machine-learning model; generating, based on the plurality of wavelet coefficients, a synthetic image of the second type; comparing the synthetic image of the second type with the training image of the second type; and updating the machine-learning model based on the comparison.

Disclosed are systems and methods including software processes for developing test cases for testing robustness of AI-based image-recognition models-under-test (MUTs) with respect to types of image variation transformations. The system may generate various types of robustness metrics for the MUT and output user-readable reports about the MUT's performance. The system trains machine-learning architectures to generate test cases including augmented images according to the types of image transformations, applies the IR MUTs, and then evaluates the image feature vector embeddings and predicted classification produced by the IR MUTs to determine the accuracy of the MUT with respect to each type of transformation.

The invention relates to a cross-domain infrared target detection method based on frequency domain style enhancement, and belongs to the technical field of infrared image target detection in computer vision. A Fourier transform domain migration module converts a source domain image and a target domain image into a frequency domain space based on Fourier transform; a similar target domain image is generated by replacing the low-frequency amplitude component of the source domain image with the low-frequency amplitude component of the target domain image; the difference between the class target domain and the target domain on the image level and the class level is reduced through the coarse-to-fine domain alignment module, and it is guaranteed that the final target detection model keeps efficient detection performance in the target domain. According to the method, the similar target domain image is constructed, the migration problem of the two sub-domains is solved in sequence, effective migration from the source domain to the target domain is achieved, different characteristics of frequency domain information and a coarse-to-fine domain alignment strategy are utilized, the influence of domain migration is reduced to the maximum extent, and therefore the accuracy and robustness of target detection are improved.

Described are systems and methods for training a machine-learning model to generate image of biological samples, and systems and methods for generating enhanced images of biological samples. The method for training a machine-learning model to generate images of biological samples may include obtaining a plurality of training images comprising a training image of a first type, and a training image of a second type. The method may also include generating, based on the training image of the first type, a plurality of wavelet coefficients using the machine-learning model; generating, based on the plurality of wavelet coefficients, a synthetic image of the second type; comparing the synthetic image of the second type with the training image of the second type; and updating the machine-learning model based on the comparison.

The invention discloses a panoramic image determination method and device, equipment, a medium and a product. The method comprises the following steps: acquiring a current scene image, shot by at least one camera device, of an environment to which a vehicle belongs, and kinematics data and attitude data of the vehicle from a current shooting moment to a previous shooting moment; determining a scene image transformation matrix based on the kinematics data and the attitude data; acquiring a historical scene image shot by the camera device at a previous shooting moment, and performing stereo matching on the current scene image and the historical scene image based on the scene image transformation matrix and shooting parameters of the camera device to obtain a scene stereo image corresponding to the camera device; and determining and displaying a panoramic image based on the scene stereo image of each camera device. The authenticity, accuracy and reliability of panoramic image determination are improved, and the driving safety is improved.

The invention relates to the technical field of image geographic positioning, in particular to a progressive cross-view-angle image geographic positioning method based on spatial feature aggregation and position awareness, which comprises the following steps: acquiring paired satellite images and ground panoramic images as sample images, and constructing a training sample set; a cross-view image geographic positioning model composed of a double-branch backbone network and a fine-grained prediction network is constructed, two branches of the double-branch backbone network are a satellite processing branch and a ground panorama processing branch, and the fine-grained prediction network is composed of a bird's-eye view image transformation module and a position sensing prediction module; inputting training samples in the training sample set into the cross-view image geographic positioning model, and performing iterative training on the cross-view image geographic positioning model until convergence; and inputting a to-be-positioned paired satellite image and ground panoramic image into the trained cross-view-angle image geographic positioning model to obtain a positioning result output by the trained cross-view-angle image geographic positioning model.

The invention provides an automatic image cutting method and system based on face key points, which are applied to the technical field of image processing, and the method comprises the steps: inputting a portrait image into a portrait segmentation model for foreground extraction, and obtaining a figure region mask; performing top positioning estimation based on the figure area mask to obtain a top position of the portrait image; inputting the portrait image into a face key point detection model to obtain a plurality of key points; determining a face midpoint of the portrait image based on the position relationship of the plurality of key points; performing affine transformation estimation based on the source point set and a preset target point set to obtain an affine transformation matrix; adjusting the vertical offset of the affine transformation matrix to obtain a longitudinal correction matrix; performing edge correction on the longitudinal correction matrix to obtain a cutting transformation matrix; and performing image transformation on the portrait image based on the cutting transformation matrix to obtain a standard composition image. According to the invention, the cut image with uniform size, standard composition and good detail retention can be generated.

A computing system may include an image access engine configured to access a panoramic point cloud image of a physical environment. The computing system may also include an environment location-aware text engine configured to transform the panoramic point cloud image into an alternate representation that reduces distortion in the panoramic point cloud image and perform an optical character recognition (OCR) process on the alternate representation to determine text in the panoramic point cloud image. The environment location-aware text engine may further be configured to construct text labels to track the text determined in the panoramic point cloud image and support text searches for the physical environment through the text labels.

The invention discloses a heterogeneous remote sensing image transformation method and system, and the method comprises the following steps: S1, image preprocessing, S2, deep multi-modal feature extraction, S3, graph theory optimization feature matching, S4, hybrid adaptive transformation model construction, S5, image fusion enhancement, and S6, perception driving quality enhancement. The heterogeneous remote sensing image transformation system comprises a multivariate data access and management module, an ultra-precision preprocessing engine module, a depth feature extraction and precise matching unit module, a self-adaptive high-precision image transformation model construction and optimization module, a refined image fusion and quality enhancement unit module and a result display and interaction module. According to the method, the problems of noise residues, radiation errors and the like in a traditional method are solved by combining technologies of ultra-precise image preprocessing, depth feature extraction and precise matching, self-adaptive high-precision image transformation model construction and refined image fusion and quality enhancement.

Systems and methods for view-conditioned diffusion for real-world vehicle gaussian splatting. A single perspective image can be transformed using image transformation techniques to generate a training dataset that addresses a domain gap between synthetic data and real-world data in a traffic scene. A pre-trained diffusion model can be finetuned with the training dataset to obtain a fine-tuned diffusion model. Perspective-aware images having different perspective views of an entity from the single perspective image can be generated using the fine-tuned diffusion model. A large generative model (LGM) can be trained using the perspective-aware images to generate a gaussian splatting model for the entity. View-conditioned simulations from the single perspective image can be generated by using the gaussian splatting model for downstream tasks.

The invention relates to a three-dimensional face recognition method for an underground coal mine, and belongs to the technical field of multi-dimensional face recognition. According to the method, the two-dimensional face image is converted into the three-dimensional points under the camera coordinate system, the rotation matrix is combined for rotation operation, the side face data of the rotating face at multiple angles are generated, and a rich data basis is provided for three-dimensional reconstruction. PointNet + + and an expansion structure thereof are combined with an auto-encoder frame to establish a three-dimensional reconstruction model, compact feature representation can be extracted from a three-dimensional point cloud, the features are utilized to reconstruct a three-dimensional face image with details and geometric structures, the three-dimensional face image is divided into a plurality of recognition areas, and feature recognition matching is performed respectively; during recognition, a face recognition result can be directly given only by screening clear areas for image reconstruction and feature matching. According to the invention, the accuracy and fidelity of reconstruction are improved.

The application discloses a similarity optimal on-board image registration method based on inertial navigation data fast convergence, and comprises the following steps: constructing a projection transformation model between an image transformation matrix, a to-be-registered image and a reference image; obtaining the projection transformation matrix according to the field of view optical axis position and three-angle offset of the satellite at the imaging moment and coarse registration; performing global coarse matching by using the projection transformation model and the projection transformation matrix to obtain a coarse matching image; decoupling the image transformation matrix according to a mathematical model to obtain a to-be-solved transformation parameter, substituting the to-be-solved transformation parameter into the image transformation matrix to obtain an optimal transformation matrix, and applying the optimal transformation matrix to the to-be-registered image to obtain a fine registration image; and calculating the similarity of the fine registration image and the reference image according to a local normalized image similarity measurement algorithm to obtain an optimal matching result. The application has the advantages of meeting the on-board calculation capacity requirement, meeting the image processing precision requirement and meeting the on-board storage resource requirement.

The invention relates to the field of visual identification, and discloses a visual identification-based confused APP automatic test system. The system comprises an image input module, an image preprocessing module, a manifold learning module, an information geometry matching module, an image conversion module, a dynamic element processing module, an automatic operation execution module and a report generation module. The invention further provides an automatic confused APP testing method based on visual identification, and the method comprises the steps of accurately positioning a control and executing automatic testing operation through technologies such as image preprocessing, manifold learning, information geometric calculation and Kalman filtering, and finally recording a result and generating a report. By combining manifold learning, information geometry, Kalman filtering and other technologies, the control positioning precision and the dynamic element adaptive capacity are improved, the image preprocessing efficiency is optimized, automatic test operation execution is achieved, the problems of inaccurate control recognition and the like in the prior art are solved, and the test stability and reliability are improved.