52 results about "Pixel value difference" patented technology

Disclosed is a method for enabling exposure-guided three-dimensional (3D) reconstruction, including (i) receiving Low Dynamic Range (LDR) input image captured by camera; (ii) rendering High Dynamic Range (HDR) image corresponding to input image, using HDR 3D reconstruction model pre-trained for HDR image reconstruction of 3D environment; (iii) applying HDR to LDR tone-mapping operator on HDR image, for producing tone-mapped LDR image; (iv) determining first loss function (FLF) between input image and tone-mapped LDR image, FLF comprising weighted pixel value differences between pixels of input image and corresponding pixels of tone-mapped LDR image; (v) determining whether input image has saturated pixel(s), saturated pixel(s) comprises: highlight saturated pixel, shadow saturated pixel; (vi) de-weighting pixel value difference corresponding to saturated pixel in FLF; (vii) back-propagating gradient of FLF with respect to model parameters, through differentiable render function of HDR 3D reconstruction model, for adjusting model parameters in way that reduces FLF.

The application provides a lamp bead arrangement method and device of an infrared touch screen and electronic equipment. The method comprises the following steps: acquiring a light net image of the infrared touch screen, and converting the light net image into a single-channel light net image; dividing the single-channel light net image into a plurality of light net areas, and acquiring pixel mean values of the plurality of light net areas; determining a difference between a maximum pixel mean value and a minimum pixel mean value in the pixel mean values of the plurality of light net areas, and judging whether the difference is greater than or equal to a preset pixel value difference threshold; in response to the difference being greater than or equal to the preset pixel value difference threshold, moving a lamp bead corresponding to a light net area corresponding to the minimum pixel mean value to a light net area corresponding to the maximum pixel mean value. The light net distribution of the infrared touch screen is more uniform.

The invention discloses a full-process visual monitoring and intelligent management method for a glass bracket production line, and relates to the technical field of industrial quality control, and the method comprises the steps: carrying out the continuous image collection of all stations of the glass bracket production line through an image collection unit, and constructing a full-process image data sequence; detecting the size of the glass bracket by using the qualified glass bracket size image database, and judging whether the size is abnormal or not; further performing defect detection on the glass bracket with the normal size, and determining whether an abnormal defect exists or not through pixel value difference analysis; and finally, summarizing the glass brackets with abnormal sizes or defects and related information thereof, and generating a visual quality report, so as to quickly position problems and optimize the production process. The comprehensive monitoring and intelligent management of the glass bracket production process are realized, and the production efficiency and the product quality are improved.

The application relates to the field of infrared thermal imaging technology, and particularly provides an infrared thermal imaging AI intelligent temperature measurement method.The method comprises the following steps: image input and preprocessing, receiving infrared images and sunlight images photographed by an infrared thermal imager, classifying based on pixel value differences, and preprocessing to obtain binary images; image cutting, cutting the binary images, and identifying the boundaries of numbers and symbols; characteristic value calculation and extraction, dividing regions according to the boundaries of the numbers and symbols, and performing feature extraction to construct a characteristic vector; number classification and identification, comparing the characteristic vector with a pre-trained K-nearest neighbor algorithm (KNN) model to identify the numbers and symbols; and image number output, combining the identified numbers and symbols into complete temperature values, outputting the identification result, and storing the identification result for comparison diagnosis.The method automatically and accurately identifies temperature information in the infrared images, and realizes intelligent comparison diagnosis.

The invention discloses a video frame rendering method, system and device, a storage medium and a program product, and belongs to the technical field of video processing. And a decoder in the CPU transmits the reference frame data to a reference frame cache region of the GPU. And a graph renderer in the GPU renders the reference frame according to the reference frame data. And a decoder in the CPU generates residual data according to the pixel value difference between the target frame and the reference frame, and transmits the residual data to a residual cache region of the GPU. And a graphic renderer in the GPU renders a local image area of the target frame according to the residual data, and superposes the local image area of the target frame with the reference frame to obtain a rendered image of the target frame. The technical effects of reducing GPU computing resources and memory bandwidth consumption, reducing data transmission delay and improving rendering efficiency and resource utilization rate in a high-resolution and high-frame-rate scene are achieved.

One embodiment of a video processing device according to the present invention comprises an acquisition unit, an extraction unit, a difference calculation unit, an accumulation unit, a time position calculation unit, a function calculation unit, and a prediction unit. The acquisition unit acquires original video representing an iterative operation in which an orientation change is repeated. The extraction unit extracts a plurality of frames from the original video. The difference calculation unit calculates a pixel value for the difference between adjacent frames among the plurality of frames. The accumulation unit accumulates the pixel values for the difference in each cycle of the iterative operation. The time position calculation unit calculates the time position of an orientation change within the cycle on the basis of the result of accumulating the pixel values for the difference. The function position calculation unit calculates a cumulative distribution function for associating each frame in the cycle with the time position of the orientation change represented by each frame. The prediction unit predicts a time position of each frame in the cycle on the basis of the cumulative distribution function.

The invention provides a radiomics feature extraction and analysis method for a liver cancer CT image, and the method comprises the steps: obtaining a multi-time-phase computed tomography image sequence, extracting contrast agent concentration distribution data of each time phase from the sequence, and obtaining a concentration change rate sequence through pixel value difference calculation; for the rapid change stage, applying a feature extraction algorithm to obtain high-density texture features from the image data of the stage, and obtaining quantitative description of subtle instantaneous change; determining the distribution position and intensity of the abnormal texture in the rapid change stage through the quantitative description, and generating an abnormal texture distribution diagram; and according to the abnormal texture distribution diagram and the overall trend summarization sequence, fusing the abnormal texture distribution diagram and the overall trend summarization sequence to generate a comprehensive dynamic change model, and determining an adjustment parameter of the analysis fineness degree of each stage.

The application discloses a kind of based on optimization's guiding filter image fusion change detection method and device, method includes: by extreme minimum scale difference operator, pixel value difference value is to the picture of same scene different phase, obtains difference map;Using the way of guiding filter, difference map is fused on global scale, to obtain the optimal difference map;Mean filter is used to phase map I1 and I2, and the base layer difference map corresponding to each difference map is obtained, and the optimized brightness saliency map F1 and F2 are obtained by pca fusion as the image of generating saliency map, according to the principle of maximum saliency to determine weight map;Weight map is regularized for double-scale difference map reconstruction, and the final difference map is obtained;In clustering segmentation stage, by introducing soft threshold function, the final difference map is further processed, to suppress existing abnormal noise.The device includes: processor and memory.

The present application relates to the technical field of multi-image analysis, in particular to a highway pavement quality detection method and system based on machine vision. According to the present application, the brightness crack probability is obtained according to the gray scale difference of the same suspected crack area in the pavement gray scale images before and after watering, and the temperature crack probability is obtained according to the pixel value difference of the same suspected crack area in the pavement infrared images before and after watering and the environmental data; according to the significant degree of the pixel value difference of the same suspected crack area in the pavement feature images before and after watering, the brightness crack probability and the temperature crack probability are adjusted to obtain the final crack probability of the suspected crack area, and the pavement quality of the to-be-tested highway is detected by using the same. The present application combines the brightness change characteristics and temperature change characteristics of the crack before and after watering, reduces the dependence on light, and improves the accuracy of pavement crack detection.

The invention discloses a video stream playing quality detection method and system based on an embedded system, and belongs to the technical field of video processing, and the method comprises the steps: carrying out the preprocessing of a target video based on Gaussian filtering, and carrying out the convolution calculation based on a mask matrix to retain the edge contour information, and obtaining a to-be-processed video. The accuracy and efficiency of subsequent processing are improved; converting video frames of the to-be-processed video into pictures, performing gray conversion, traversing pixels of gray images, and determining whether continuous video frames are lagged video frames or not according to pixel value differences; processing the video frame through gray level conversion and Canny edge change to obtain an edge feature map, and performing matching based on a preset mosaic template; performing cross weighting processing on the matched area to confirm existence of artificial mosaics and non-artificial mosaics; according to the lag data and the non-artificial mosaic data, the playing quality is determined, and the playing quality of the video can be reasonably identified and judged.

The invention relates to the technical field of image processing, in particular to a line etching method and device for a thermal printing sheet, and the method comprises the steps: obtaining each structural region in an exposure image of the thermal printing sheet, carrying out the equal division of all pixel values in each structural region, and analyzing the pixel order change degree and the pixel value difference in the region, acquiring the pixel disorder degree of each structural area; the method comprises the following steps of: obtaining a plurality of structural areas, obtaining the angular point distribution difference degree of each structural area by utilizing the angular point characteristics in each structural area, taking the corresponding structural areas with the same angular point distribution difference degree as the same structural area of each structure, and further obtaining the search step length of the same structural area of each structure so as to carry out line etching on the thermal printing sheet. The circuit etching efficiency of the thermal printing sheet can be improved.

The invention discloses an intelligent cloth printing defect monitoring method and system based on machine vision, and relates to the field of image data processing, and the method comprises the steps: collecting a real-time video stream on the surface of a fabric, intercepting single-frame image data, and carrying out the grid segmentation, so as to obtain candidate image blocks; performing feature clustering analysis on sample blocks in the candidate image blocks, and constructing a standard texture feature library; sparse coding operation is carried out on the candidate image blocks based on the standard texture feature library, linear combination weights are solved, and sparse coefficients are obtained; generating a reconstructed image block according to the standard texture feature library and the sparse coefficient; calculating a pixel value difference between the candidate image block and the reconstructed image block to obtain a residual matrix representing a difference degree; and when the energy norm value of the residual matrix is greater than an anomaly judgment threshold value, outputting a marking signal indicating that the corresponding candidate image block has an image defect. By implementing the method and the device, defect detection without a preset template can be realized, and the adaptability of printing defect monitoring is improved.

The application discloses a pattern defect detection method and device and a storage medium. The method comprises the following steps: obtaining an overall pattern region and a plurality of local pattern regions corresponding to a target pattern in a to-be-detected image, and obtaining an overall template image and a plurality of local template images from template information corresponding to the target pattern; aligning the overall pattern region with the overall template image, and respectively aligning each local pattern region with a corresponding local template image; determining a first defect position in the overall pattern region based on a first pixel value difference between the aligned overall pattern region and the overall template image, and respectively determining a second defect position in each local pattern region based on a second pixel value difference between the aligned each local pattern region and the corresponding local template image; and comprehensively determining a third defect position of the target pattern in the to-be-detected image based on the first defect position and the second defect positions. Through the above method, the accuracy of pattern defect detection can be improved.

The application discloses a lossless compression method and system for image sensor raw data, and relates to the technical field of image sensors.The application fully utilizes the statistical characteristics that the pixel value difference between adjacent frames under the same exposure is extremely small, combines the inherent color arrangement rule of the Bayer format image, separates the color channels first and then performs time domain difference calculation, greatly reduces the data entropy value, and obtains a compression ratio much higher than that of a general lossless compression algorithm on the premise of completely lossless original pixel values; meanwhile, independent compression and packaging are performed with the minimum service unit as the granularity, target data can be directly located and accessed without decompression and large-scale package, the data retrieval and calculation overheads are reduced, the traceability of the original data is retained, the statistical confidence of the test result is not affected, and the batch data storage requirement in the image sensor production test scene is adapted.

The invention relates to the technical field of laser cutting, in particular to a metal plate laser cutting machining device and a laser cutting machining method thereof. The method comprises the following steps: selecting a puncture point of a contour, and starting cutting at the puncture point; according to the pixel value difference of corresponding pixel points in the sheet metal infrared image of the start cutting point of the cutting crack in the local time period at the start and end moments, the position and the pixel value of the pixel points on the cutting crack, and the cutting position and the width of the cutting crack, the thermal response goodness degree is obtained; according to the relative size of the puncture preferences of pixel points on cutting cracks of each local time period and the previous local time period of each contour and the thermal response optimization degree of the previous local time period, the pulse demand degree is obtained, then the actual pulse frequency of the previous local time period is adjusted, the optimal pulse frequency of each local time period is obtained, and the optimal pulse frequency of each local time period is obtained; and cutting of each contour is completed. According to the method, the optimal pulse frequency of the local time period of each contour is determined in a self-adaptive mode, and the metal plate laser cutting precision is improved.

This invention relates to the field of data anomaly detection technology, and more particularly to a method for detecting abnormal behavior in surveillance video data based on deep learning. The method acquires video images of any individual in real time and detects them. When no frontal position information of the individual is detected, the video image is marked as a target image. Multiple consecutive frames of target images are acquired to obtain a target image set. Based on the head and hand position information of any individual in each frame of the target image set, motion coefficients are obtained to identify whether any individual exhibits suspected abnormal behavior. If any individual exhibits suspected abnormal behavior, the abnormal behavior detection result is determined based on the smoke concentration data in the target area and the pixel value differences of pixels in each frame of the target image set. This invention can accurately identify abnormal behavior of employees in a target area.

The invention relates to the technical field of image area detection, in particular to a tea wall breaking rate detection method, equipment and system. The method comprises the following steps: acquiring gray images of leaves at different angles after each batch of rolling, and initially quantifying the wall breaking condition through a pixel value difference area to obtain an initial wall breaking index; determining a suspected crack edge according to the edge direction consistency degree and the width smoothness condition, obtaining a wall breaking sufficient evaluation index according to the length position distribution uniformity condition of the suspected crack, and analyzing the boundary complexity to adjust the initial index to obtain an adjusted wall breaking index; and obtaining a confidence coefficient by adjusting index consistency and a historical batch similarity degree through different angles, and determining a final wall breaking rate based on the confidence coefficient. According to the method, wall breaking evaluation is adjusted by considering the influence of leaf veins and tearing characteristics, abnormal conditions caused by process drift are found in time through real-time detection and correlation historical process reference analysis after rolling, and stable quality of tea production is guaranteed while the accuracy of wall breaking evaluation is improved in factory production.

An object of the present invention is to solve the problems of the prior art, that is, to provide a method for determining the quality of a placing joint surface treatment, which can objectively determine the quality of the placing joint surface treatment after estimating the distribution state of aggregates on a concrete surface.SOLUTION: The construction joint surface processing determination method is a method for determining the quality of construction joint surface processing on the basis of an original image obtained by photographing a concrete surface, and includes an original image acquisition process and an aggregate extraction process. In the original image acquisition step, a first smoothed image and a second smoothed image are generated by smoothing an original image. In the aggregate extraction step, a portion where a pixel value difference between the first smoothed image and the second smoothed image exceeds an aggregate threshold value is extracted as an aggregate portion. To determine the quality of placing joint surface processing on the basis of the ratio of an aggregate part occupying an object area corresponding to an original image.SELECTED DRAWING: Figure 3

A thermal image processing method according to the present invention comprises the steps of: receiving a captured image from a thermal image camera; dividing each frame included in the captured image into a plurality of blocks; identifying each block as a target block or an excluded block on the basis of the pixel value difference between corresponding blocks of two neighboring frames and / or the average pixel value of each block of the frame; determining whether to mark each block position on the basis of the cumulative count of the target block within a plurality of frames; calculating a non-uniformity correction value for an unmarked block position; and outputting a corrected image by correcting the captured image on the basis of the calculated non-uniformity correction value.

The application relates to a defect identification method, device and computer equipment, which comprises the following steps: acquiring an image containing a power small fitting collected by an image collection device; determining a depth threshold according to an actual shooting distance of the image; wherein the actual shooting distance is the actual distance between the image collection device and the power small fitting, and the depth threshold is positively correlated with the actual shooting distance; extracting a region of interest containing the power small fitting from the image according to the depth threshold; determining the definition of the region of interest according to the pixel value difference of different pixel points in the region of interest; and performing defect identification of the power small fitting according to the region of interest and the definition when the definition is greater than a preset definition threshold. The application can improve the accuracy and efficiency of image screening to ensure the accuracy and efficiency of defect identification.

An imaging system includes a controllable light source; an image sensor; metalens for focusing incoming light onto the image sensor; and processor(s). The processor(s) is configured to control a light source using a first illumination intensity and / or first illumination wavelength, while controlling the image sensor to capture first image; control the light source using a second illumination intensity and / or second illumination wavelength, while controlling the image sensor to capture a second image; calculate measured differences between pixel values of pixels in the first image and pixel values of corresponding pixels in the second image; estimate expected pixel value differences based on difference between the first and second illumination intensities and / or differences between the first and second illumination wavelengths; and correct pixel values of pixels in the first image and / or second image based on deviation in measured differences from expected differences.

The invention discloses an image magnification method, an image magnification device, an image processing chip and display equipment. The method comprises the following steps: receiving an original image; carrying out edge detection on the original image and calculating a pixel value difference value in the original image; judging whether the pixel value difference value is greater than a preset threshold value or not; when the pixel value difference value is greater than a preset threshold value, amplifying the original image by adopting a first interpolation method; and when the pixel value difference value is not greater than a preset threshold value, amplifying the original image by adopting a second interpolation method. According to the image magnification method, the image magnification device, the image processing chip and the display equipment provided by the invention, the edge part of the image is detected, and different magnification algorithms are selected according to the pixel value difference, so that the display effect of a color image is reserved, and the sharpness of images such as an image engineering drawing and fonts after magnification is enhanced.

In the present invention, an information processing device is capable of generating sub-images to be sequentially displayed, and comprises: a first processing unit that, on the basis of a frame image, is capable of generating two or more sub-images among a plurality of sub-images having a plurality of pixel values and arranged in a predetermined arrangement pattern in the frame image; a detection unit that detects detection pixels to be processed, on the basis of the display order of corresponding pixels of the two or more sub-images and the pixel value difference between the pixels in the display order and on the basis of the coordinates of the frame image; and a second processing unit that executes processing for reducing the pixel value difference between corresponding pixels in the detection pixels.

The application provides an image frame encoding method and device, a chip, electronic equipment, a storage medium and a computer program product, and relates to the field of image processing. The method comprises the following steps: obtaining a current image frame, the current image frame comprising a first image frame and a second image frame, the first image frame being a previous image frame of the current image frame, the second image frame being a previous image frame of the first image frame, and the current image frame being an image frame in a video stream that needs to be transmitted from a first node to a second node; determining a current predicted image frame according to the first image frame and the second image frame; determining a residual image between the current image frame and the current predicted image frame; determining a plurality of first image blocks with pixel value differences between the current image frame and the first image frame; in a case where the number of the first image blocks is less than or equal to a preset number, losslessly compressing image blocks in the residual image that are located at the same positions as the plurality of first image blocks; and transmitting the losslessly compressed image blocks to the second node.

This invention provides an intelligent method for restoring hidden targets in remote sensing images based on image inpainting. It constructs a gated convolutional target restoration network and trains an image inpainting model capable of adapting to remote sensing images with significant color variations, fine granularity, and complex features. The gating mechanism provides remote sensing target mask information in the front-end network, while the back-end network acts as an attention mechanism, dynamically learning the pixel distribution characteristics of uncovered target areas and non-target areas in the remote sensing image. This invention uses the original remote sensing image, a combination of a random overlay mask and a target contour mask as input for model training. Guided by labels indicating the hidden target area and the exact location of the remote sensing target, the model can quickly identify key targets in remote sensing images with complex color gradients and large pixel value differences. It then learns the contour features of the remote sensing target and the complex features of its surrounding environment, ensuring that the restored remote sensing target conforms to the features and distribution patterns of ground features in the original image.

The present disclosure relates to the technical field of image processing, and discloses an image feature extraction method, device, equipment and storage medium. The method comprises: acquiring an image to be extracted; determining a pixel value difference result of each pixel and at least one surrounding pixel based on the image to generate a difference matrix; performing vectorization processing on the difference matrix to generate a difference vector; and performing binaryzation processing on the difference vector to generate a feature vector corresponding to the image. Through the above technical solution, the feature of the image is extracted by using image spatial domain difference transformation, the calculation complexity and amount of feature extraction are reduced, and the information comprehensiveness and efficiency of feature extraction are improved.

The present application relates to the technical field of laser cutting, and particularly relates to a sheet metal laser cutting device and a laser cutting method thereof. The present application selects a piercing point of an outline and starts cutting at the piercing point; according to the pixel value difference of the corresponding pixel points in the sheet metal infrared image of the start and end time of the starting cutting point of the cutting crack in the local time period, the position and pixel value of the pixel points on the cutting crack, and the cutting position and the cutting crack width, the heat response position optimization degree is obtained; according to the relative size of the piercing optimization degree of the pixel points on the cutting crack of each local time period and the previous local time period of each outline, and the heat response position optimization degree of the previous local time period, the pulse demand degree is obtained, and then the actual pulse frequency of the previous local time period is adjusted, the preferred pulse frequency of each local time period is obtained, and the cutting of each outline is completed. The present application adaptively determines the preferred pulse frequency of the local time period of each outline, and improves the sheet metal laser cutting precision.

This invention discloses an underwater image collaborative enhancement and super-resolution method, belonging to the field of computer vision technology. It includes constructing a training dataset for the model training process using an underwater image dataset with visual enhancement reference images. The training dataset includes low-resolution original underwater degraded images, low-resolution reference ground truth images, high-resolution original degraded underwater images, and high-resolution reference ground truth images. An underwater image collaborative enhancement and super-resolution deep learning network is constructed, using pairs of original underwater degraded images as input to the deep network, and using their corresponding low-resolution reference ground truth images to constrain and optimize the predicted images output by the model. The model training process is constrained and optimized by minimizing pixel value difference loss and structural similarity loss, thereby training the underwater image collaborative enhancement and super-resolution deep learning network. Any pair of original underwater degraded images is used as input to the model, and the output is obtained through model calculation.

The invention relates to the technical field of CL imaging, in particular to a CL system rotating shaft projection position calculation method. The method comprises the steps that a sample is placed on an objective table, the height is adjusted to ensure complete projection capture, the objective table is controlled to rotate by a circle to collect a plurality of projection images with marked angles and coordinates at equal angle intervals, all the projections are superposed to generate a composite image, and fixed feature points form an elliptical orbit; delimiting a row coordinate interval containing effective sample projection points in the row direction of the composite image as a candidate symmetry axis search range, traversing the row position of each hypothetical symmetry axis in the range, constructing a symmetric row pair combination, and calculating the sum of pixel value differences of all row pairs under each hypothetical symmetry axis; and selecting the row coordinate with the minimum difference sum as the projection position of the rotating shaft in the Y direction of the detector. According to the method, high-precision positioning is realized through symmetry analysis of the projection superposition image, a die body does not need to be calibrated, the stability of a calculation result is remarkably superior to that of a traditional technology, and the definition of a reconstructed image is effectively improved.

Disclosed is a method for enabling exposure-guided three-dimensional (3D) reconstruction, including (i) receiving Low Dynamic Range (LDR) input image captured by camera; (ii) rendering High Dynamic Range (HDR) image corresponding to input image, using HDR 3D reconstruction model pre-trained for HDR image reconstruction of 3D environment; (iii) applying HDR to LDR tone-mapping operator on HDR image, for producing tone-mapped LDR image; (iv) determining first loss function (FLF) between input image and tone-mapped LDR image, FLF comprising weighted pixel value differences between pixels of input image and corresponding pixels of tone-mapped LDR image; (v) determining whether input image has saturated pixel(s), saturated pixel(s) comprises: highlight saturated pixel, shadow saturated pixel; (vi) de-weighting pixel value difference corresponding to saturated pixel in FLF; (vii) back-propagating gradient of FLF with respect to model parameters, through differentiable render function of HDR 3D reconstruction model, for adjusting model parameters in way that reduces FLF.