25 results about "Histogram matching" patented technology

To employ low resolution, noisy tone mapping operations for high resolution images, at least one raw image frame is converted to a first image at a higher resolution and a second image at a lower resolution. Tone mapping is applied to the second image to derive a third image at the lower resolution. Histogram matching and regularization are performed to determine a lookup table approximating histogram matching of the second image to the third image. A global gain map is derived based on luma for the first image and the lookup table. A local gain map is derived by up-sampling and denoising residual differences between the third image and the lookup table applied to the second image. Based on the global gain map and the local gain map, a total gain map is determined for tone mapping the first image to produce a fourth image at the first resolution.

A method for detecting defect in image is provided. The method obtains a number of original images, determines a first reference image from the original images, and performs a histogram matching on the original images excluding the first reference image according to the first reference image, to obtain a plurality of matched images. The method further generates a synthesized image according to pixel intensities of the matched images and pixel intensities of the first reference image; and uses the synthesized image as a second reference image to perform an image comparison with a test image, to generate a result of defect detection. A related device and a related non-transitory storage medium are also provided.

The invention belongs to the field of panoramic stitching digital image processing technologies and camera ISP technologies, and particularly relates to a chromatic aberration consistency control method based on front and rear end collaboration. The method comprises the following steps: 1, acquiring statistical information of multiple cameras; step 2, acquiring a current camera control parameter by adopting a front-end chromatic aberration consistency adjustment algorithm; 3, the current camera control parameters are issued to all the cameras, and all the cameras complete exposure imaging according to the current camera control parameters; step 4, acquiring an overlapped part image in each path of camera imaging; step 5, calculating a statistical histogram cumulative distribution function of overlapped part images in each path of camera imaging; step 6, calculating a histogram matching function according to the cumulative distribution function of the overlapped part images in each path of overlapped camera imaging, so that the histograms of the overlapped parts of the two images tend to be consistent; and step 7, according to the obtained histogram matching function, transforming the brightness channel of each pixel in the corresponding camera imaging to complete color difference consistency correction.

The application discloses a backdoor attack method based on color frequency injection and adaptive local enhancement, and relates to the technical field of machine learning and artificial intelligence security. The method comprises the following steps: introducing low-frequency color offset and weak high-frequency signal into an image in a CIELAB color space to perform global color-frequency injection; using a pre-trained proxy model to locate a high-sensitive perception domain of the model through mixed evaluation of gradients and class activation maps, and generating a binary mask; in an HSV color space, respectively applying nonlinear stretching factors to saturation and brightness of the sensitive domain based on the mask to perform adaptive local enhancement; using Gaussian smoothing, adaptive noise and histogram matching to eliminate edges and statistical abnormalities caused by local enhancement, completing compensation color enhancement to generate a poisoned image; and modeling a trigger core parameter as a constrained optimization problem, and using a particle swarm optimization algorithm to jointly dynamically update the trigger core parameter to obtain an optimal strategy. The application anchors the trigger feature depth in the core semantic area of the model and lurks in the normal data manifold, guarantees a high attack success rate, realizes extreme visual and feature concealment, and has strong anti-defense robustness.

The invention discloses a satellite remote sensing image fusion method based on Wrapping-Curvelet transformation, and the method comprises the steps: carrying out the spatial registration of a multispectral image and a panchromatic image, carrying out the resampling of the multispectral image according to bilinear interpolation, obtaining a multispectral image with the same pixel size as the panchromatic image, and carrying out the fusion of the multispectral image and the panchromatic image. The method comprises the following steps of: respectively performing histogram matching on a panchromatic image and three wave bands of a multispectral image to be fused to obtain a new panchromatic image, taking the three wave bands of the multispectral image and the corresponding panchromatic image as examples, respectively performing Wrapping-Curvelet transformation on the three wave bands of the multispectral image and the corresponding panchromatic image to obtain respective Curvelet coefficients, fusing the coefficients of each layer, and then performing inverse Wrapping-Curvelet transformation on the fused layers to obtain a new panchromatic image; and obtaining a fused image. The blocking effect caused by blocking is avoided, the number of parameters of the implementation method is small, redundancy is reduced, and the algorithm is simple, convenient and rapid.

The application provides a style transfer strong similarity-based generation method, device, equipment and medium, the method comprising: S10, redrawing a style reference map, outputting a clear style map; S20, inputting the clear style map into a monocular depth estimation model, outputting a single-channel depth map; S30, inputting the single-channel depth map into an enhanced image generation model for structure optimization, generating a redrawing sketch; S40, using a visual feature extraction model and a deep learning model to perform style fusion and rendering on the redrawing sketch, generating a preliminary style map; S50, pre-training a LoRA model, using the LoRA model to adjust the preliminary style map, outputting an intermediate style map; S60, after upsampling the intermediate style map, performing block optimization, outputting an optimized style map; S70, inputting the optimized style map into a Lab color space, performing histogram matching, and generating a final style map.

The application discloses a kind of micro-light image enhancement methods, including obtaining target micro-light image;Based on gamma correction, RTV correction and histogram matching scheme, the structure diagram and texture diagram of target micro-light image are calculated;According to the structure diagram and texture diagram obtained, based on Retinex model and structure / texture guide scheme, the brightness diagram and reflection diagram of target micro-light image are calculated;The obtained brightness diagram is adjusted and adaptively corrected;Based on RTV correction scheme, the enhanced image of target micro-light image is calculated to complete the image enhancement of target micro-light image.The application also discloses a kind of mine image enhancement method comprising the micro-light image enhancement method.The application not only realizes the enhancement of micro-light image and mine image, but also has higher reliability, better accuracy and better enhancement effect.

The invention relates to the technical field of biological detection, and discloses an intestinal digestion flora sampling and classification detection method and system, and the method comprises the steps: analyzing a feature image, extracting high-frequency texture features and low-frequency texture features, obtaining a frequency domain feature parameter set, extracting an RGB color histogram from an original flora microscopic image, and obtaining a frequency domain feature parameter set; obtaining a color feature parameter set; according to the method, self-adaptive double-threshold edge fusion and three-dimensional skeleton topology analysis are adopted, so that the quantification precision of morphological characteristics is greatly improved, thallus contour fracture and measurement errors are reduced, and the reliability of a result is improved; secondly, the dynamic adaptability of fractal feature extraction is improved, the classification requirements of new flora can be effectively met, and the classification accuracy is improved; besides, through HSV channel strengthening and multi-channel histogram matching, color feature quantification is objectified, subjective errors in traditional manual interpretation are eliminated, and color feature recognition is more accurate.

The application discloses a high-resolution satellite image urban area shadow-removed ARD production method and device, and belongs to the technical field of remote sensing image processing. The method comprises the following steps: firstly, screening out images covering urban areas; performing radiation calibration on high-resolution satellite images covering urban areas to obtain apparent radiance images; based on the apparent radiance images, performing preliminary shadow detection by using a pre-trained deep learning model; using a near-infrared band of an atmospheric-corrected ground reflectivity image, performing false detection elimination on the preliminary detection result by adaptive threshold segmentation to obtain a final shadow mask; performing histogram matching processing on the shadow area based on the shadow mask on the atmospheric-corrected ground reflectivity image to realize shadow removal; and finally, outputting the shadow-removed ARD product. The application significantly improves the shadow detection precision and the quantitative reliability of the ARD product, and is suitable for large-scale ARD production of high-resolution satellite images in urban areas.

The application discloses a kind of based on 4D millimeter wave radar / inertial sensor tightly coupled navigation positioning method, comprising: the original point cloud is based on the dynamic point rejection of Doppler consistency, obtain first point cloud, then carry out noise point rejection and obtain second point cloud;Weighted Doppler velocity factor residual corresponding to single frame point cloud is constructed based on Doppler velocity weight;Adjacent two frames of second point cloud are histogram matched, obtain interframe point pair set, and interframe point pair factor residual is constructed based on interframe point pair set;Joint weighted Doppler velocity factor residual, interframe point pair factor residual and inertial pre-integral factor residual, construct tightly coupled nonlinear factor graph optimization model, and the navigation positioning information of carrier is obtained by executing nonlinear factor graph optimization.The application is applied to navigation positioning field, faces extreme dynamic scene, the characteristics of dynamic, static target are jointly analyzed, and positioning precision and robustness in dynamic scene are significantly improved.

The application discloses a kind of mass remote sensing image automatic color correction methods of sub-meter satellite.Belongs to the technical field of mass sub-meter satellite remote sensing image automatic color correction, specifically relates to a kind of mass remote sensing image automatic color correction methods of sub-meter satellite technical field, by training improved ResNet50 model, obtain mass sub-meter high-resolution paired data, then combined with histogram matching method, realize mass sub-meter satellite remote sensing image automatic color correction.The method comprises the following steps: obtaining pre-correction sub-meter remote sensing image, and performing histogram matching on pre-correction sub-meter remote sensing image;Pre-process the pre-correction sub-meter remote sensing image and corresponding RGB histogram;ResNet50 model is constructed and improved: after the full connection layer of ResNet50 model, increase full connection module;Sub-meter remote sensing image is corrected by ResNet50 improved model.

This application provides an anti-interference texture enhancement localization method, system, and medium suitable for low-texture visual scenes. The method includes: acquiring an original image; preprocessing the original image; extracting the texture response of the preprocessed image in different directions and frequency dimensions based on multi-scale Gabor filtering to obtain multiple texture response maps; dividing the preprocessed image into multiple image partitions based on a partitioned LED array and calculating the local gradient contrast of each image partition; independently adjusting the light source brightness of different image partitions to obtain an illumination-optimized image; performing feature fusion processing on the multiple texture response maps and the illumination-optimized image to extract feature points for visual localization; improving the consistency of images acquired at different time periods by normalizing the frequency response of the original image, histogram matching, and adaptive brightness restoration; and improving the local contrast by using a partitioned LED array with spatial illumination modulation to improve visual localization accuracy.

This invention discloses an adaptive enhancement method for low-light images based on physical prior guidance and spatiotemporal decoupling, belonging to the fields of computer vision and autonomous driving perception technology. The method first converts the original low-light image to YUV space to extract the brightness channel and performs inverse mapping. A pseudo-dehazing physical model is used to inversely solve for the transmittance map as the structural prior matrix. Subsequently, a mask decomposition network decouples the original image features into a reflection component containing local texture and an initial illumination component containing global brightness distribution. This invention constructs a dual-branch parallel enhancement architecture: the first branch uses an enhanced Transformer network for detail restoration and introduces a color histogram matching loss function to ensure color fidelity; the second branch uses an adaptive Gamma U-Net to predict pixel-level gamma parameter matrices for brightness compensation and suppresses overexposure through a regularized loss function with a range penalty term. Finally, the enhanced image is reconstructed based on Retinex theory.

Systems and methods for mitigating shadow segments from images are provided. One example computer-implemented method includes accessing an original image of a geospatial location including a first shadow segment and generating, using a model architecture, a matte for the original image. The method also includes generating a first histogram of tones of shadow pixels of the first shadow segment in the original image, generating a second histogram of tones of an adjacent region of the original image, which is proximate to the first shadow segment yet outside a boundary of the first shadow segment, defining a lookup table based on histogram matching between the first histogram and the second histogram, and relighting the original image based on the lookup table.

The invention provides an infrared image enhancement method and device based on histogram matching and curvelet transformation. The method comprises the following steps: acquiring a to-be-enhanced infrared image and a reference image; respectively calculating a gray average value and a gray standard deviation of the infrared image and the reference image; calculating a correction coefficient and an average correction value based on the gray average value and the gray standard deviation; generating a histogram matching enhanced image of the infrared image based on the correction coefficient and the average correction value; curvelet transform decomposition is carried out on the histogram matching enhanced image and the infrared image to obtain corresponding multi-scale and multi-direction sub-band coefficients, and fusion is carried out on the corresponding sub-band coefficients under the same scale and the same direction to obtain a fusion sub-band coefficient; and performing curvelet inverse transformation on the fusion sub-band coefficient to obtain a fusion enhanced image. According to the invention, through the cooperation of the global gray correction of histogram matching and the local detail fusion of curvelet transformation, the contrast, definition and detail information of the infrared image are synchronously improved.

The invention discloses a tight coupling navigation positioning method based on a 4D millimeter wave radar / inertial sensor, and the method comprises the steps: carrying out the Doppler consistency-based dynamic point elimination of an original point cloud, obtaining a first point cloud, and carrying out the noise elimination to obtain a second point cloud; constructing a weighted Doppler velocity factor residual error corresponding to the single-frame point cloud based on the Doppler velocity weight; performing histogram matching on two adjacent frames of second point clouds to obtain an inter-frame point pair set, and constructing an inter-frame point pair factor residual based on the inter-frame point pair set; and combining the weighted Doppler velocity factor residual error, the inter-frame point pair factor residual error and the inertia pre-integration factor residual error to construct a tight coupling nonlinear factor graph optimization model, and executing nonlinear factor graph optimization to obtain the navigation positioning information of the carrier. The method is applied to the field of navigation positioning, is oriented to extreme dynamic scenes, jointly analyzes the characteristics of dynamic and static targets, and remarkably improves the positioning precision and robustness in the dynamic scenes.

The application relates to the technical field of image processing, in particular to a multi-source ecological remote sensing data fusion method and system. The method comprises the following steps: collecting multi-band images and panchromatic images; obtaining a closed operation image after edge detection and closed operation of the images, and dividing the images into a plurality of partitions based on the closed operation image; respectively performing interpolation on different layer images of the panchromatic images, and correcting the interpolation based on the gradient deflection direction to obtain interpolation correction values; obtaining principal component images of the panchromatic images after the interpolation is completed, performing histogram matching on the principal component images and the panchromatic images, and adjusting the matching values based on the gray scale change and the quantity change to obtain matching adjustment values; and comparing the partition images after the matching with the multi-band images and the panchromatic images to obtain fusion values of edge pixel points, and obtaining a fusion image based on the fusion values. The application improves the definition and color performance of the fusion image.

The application discloses a kind of based on clustering and multi-scale histogram matching image sequence flicker elimination method and system, it is related to image processing technical field, method includes: using non-local mean denoising algorithm to execute denoising processing to image and obtain denoised image sequence, then calculate histogram feature, using K-Means clustering algorithm to group image in denoised image sequence, to determine the class label of each frame image;For denoised image sequence, image with class label as standard class is regarded as reference, multi-scale histogram matching is executed to image with class label as abnormal class, and all scales under optimization image are weighted fusion by Gaussian kernel function to obtain optimized image sequence;For optimized image sequence, cross-fade technology is used to carry out time sequence smooth transition processing at the junction of image of different class labels, to obtain final image sequence.The application can effectively eliminate the flicker of image sequence, and improve visual experience.

The application provides a finished product camera video real-time splicing method and system based on heterogeneous computing, which comprises the following steps: processing double-channel fisheye video streams through offline calibration to achieve global coarse alignment, and obtaining double-channel equirectangular projection video images containing target video images and reference video images; performing feature matching and inter-frame jitter suppression processing based on optical flow estimation on the target video images and the reference video images to obtain a pixel set matching point set; sequentially performing local homography matrix solving based on grid division and image warping processing on the pixel set matching point set to obtain a warped image; performing image fusion on the warped image and the reference video image through brightness matching and distance weight processing to obtain a fused image; and obtaining a fused video stream through the fused image. The application realizes pixel-level weighted fusion by defining the range through an overlapping area mask, correcting color brightness deviation through histogram matching, and generating dynamic weight coefficients based on distance transformation.

The invention discloses a shadow-removing ARD production method and device for a high-resolution satellite image urban area, and belongs to the technical field of remote sensing image processing. The method comprises the following steps: firstly, screening out an image covering an urban area; performing radiometric calibration on the high-resolution satellite image covering the urban area to obtain an apparent radiance image; based on the apparent radiance image, performing preliminary shadow detection by using a pre-trained deep learning model; performing false detection elimination on the preliminary detection result through adaptive threshold segmentation by using the near-infrared band of the surface reflectance image after atmospheric correction to obtain a final shadow mask; on the surface reflectance image after atmospheric correction, performing histogram matching processing on a shadow area based on the shadow mask to realize shadow removal; and finally, outputting the shadow-removed ARD product. According to the method, the shadow detection precision and the quantification reliability of ARD products are remarkably improved, and the method is suitable for large-scale ARD production of urban high-resolution satellite images.

The present invention relates to a method for generating diagnostic support information for axial spondyloarthritis. The method includes preprocessing an input MRI image through normalization and histogram matching. A first slice feature representing visual characteristics of the MRI image is extracted for each slice of the preprocessed MRI image by applying a first feature extraction model, and a second slice feature representing quantitative characteristics of bone marrow edema (BME) is extracted by applying a second feature extraction model to BME data associated with the MRI image. Slice-level integrated feature vectors are generated by combining the first and second slice features on a slice-by-slice basis and applying attention-based importance weights, and a patient-level final integrated feature vector is obtained by summing the weighted vectors. Diagnostic support information indicating whether axial spondyloarthritis is present is then generated by inputting the final integrated feature vector into a classification model.

The invention discloses an SAR image domain false target generation method based on time-frequency rotation, and the method comprises the steps: carrying out the local processing on the 1A-level SLC image data of an SAR, embedding a false target template, and controlling the defocusing degree of a false target in the azimuth direction through the fractional Fourier transform. Fine generation of deception jamming is achieved through fusion of a false target and an image background, it is guaranteed that the amplitude distribution of the false target is consistent with that of a real image through amplitude histogram matching, meanwhile, target phase features are reserved, and the false target close to a real target in the geometrical shape, the defocusing degree and the statistical property is generated in an SAR image. According to the method, histogram matching is carried out by fusing the image background frequency spectrum and utilizing the real target amplitude statistical characteristics, so that the problems that the difference between the false target image after defocusing processing and the real target amplitude characteristics is large and the background splitting is obvious are effectively avoided, and the fineness and the sense of reality of the false target in the image are improved.

The application discloses a face makeup transfer method based on an improved Diffusion model, comprising: preliminary data processing, obtaining a pseudo makeup dataset, the dataset is made by using the public MT dataset, the LADN dataset and the Wild-MT dataset by using histogram matching combined with a thin plate spline interpolation method; training the improved Diffusion model by using the obtained dataset to obtain an optimal makeup transfer model; the loss function of model training comprises a latent diffusion loss, a makeup loss and a diffusion reconstruction loss, the latent diffusion loss is used for guiding the model to learn to gradually restore a clear face makeup transfer image from a noise image, the makeup loss is used for optimizing the makeup effect of the generated image, and the diffusion reconstruction loss is used for keeping the consistency of the image; the optimal makeup transfer model is applied to process input source images and reference images to generate makeup transfer results with natural transition. The application can realize a high-quality and controllable makeup transfer effect while keeping the identity features of the source images.

The invention provides a gradient constraint and low tube rank regularization-based panchromatic sharpening method for a remote sensing image, which comprises the following steps of: carrying out up-sampling on a low-resolution multispectral image, carrying out histogram matching and fuzzy processing on a panchromatic image to obtain an up-sampled multispectral image, matching the panchromatic image with a fuzzy panchromatic image, and carrying out gradient constraint and low tube rank regularization on the panchromatic image to obtain the panchromatic sharpening image. Processing based on a gradient constraint mechanism to obtain a gradient constraint term; taking the up-sampling multispectral image as an initial high-resolution multispectral image, and obtaining a low-tube-rank regularization item through low-tube-rank regularization processing; and performing iterative solution based on an augmented Lagrange function and an alternating direction multiplier method, and alternately updating variables until convergence to obtain a final high-resolution multispectral image. According to the method, a variational optimization model is efficiently solved by adopting an alternating direction multiplier method, a complex problem is decomposed into a plurality of sub-problems for alternate optimization, and the convergence and real-time performance of the algorithm are ensured.

The application provides an infrared image enhancement method and device based on histogram matching and curvelet transform. The method comprises: acquiring an infrared image to be enhanced and a reference image; calculating the gray mean value and the gray standard deviation of the infrared image and the reference image respectively; calculating a correction coefficient and an average correction amount based on the gray mean value and the gray standard deviation; generating a histogram matching enhanced image of the infrared image based on the correction coefficient and the average correction amount; respectively performing curvelet transform decomposition on the histogram matching enhanced image and the infrared image to obtain corresponding multi-scale and multi-direction sub-band coefficients, and fusing corresponding sub-band coefficients at the same scale and the same direction to obtain fused sub-band coefficients; and performing curvelet inverse transform on the fused sub-band coefficients to obtain a fused enhanced image. Through the cooperation of the global gray correction of histogram matching and the local detail fusion of curvelet transform, the application realizes the synchronous improvement of the contrast, the definition and the detail information of the infrared image.