132 results about "Thresholding algorithm" patented technology

The invention discloses a method and system for identifying tunnel lining cracks based on a gradient inter-class threshold algorithm. The method includes (1) denoising the tunnel crack image; The image is binarized, and edge detection is performed according to the obtained binarized image, and the crack index is calibrated; (3) the binarized image is morphologically processed to remove the noise of the crack itself and fill the internal cavity; according to The image after filling carries out the crack indicator extraction; Its system comprises sequentially connected and comprises image enhancement module, edge detection module and crack recognition extraction module; The tunnel lining crack recognition method and system based on the gradient inter-class threshold algorithm provided by the present invention are relatively large in terms of processing capacity Huge, multi-interference images of tunnel lining cracks have a good recognition effect; it is suitable for automatic detection of lining in the skylight period of high-speed railway tunnels.

The invention discloses an SAR (Synthesized Aperture Radar) image target detection method based on zone markers and grey statistics. An artificial target potential zone thoroughly including all the artificial targets in the original SAR image is detected when independent of prior information such as target shapes, target characteristics and background characteristics and the like by using the structural information of the original SAR image extracted from a Primal Sketch model, the grey information of the original SAR image and the characteristic that the pixels of the SAR artificial target zone are distributed in an alternately dark and bright way are made the best of, a false alarm target zone is effectively removed in combination with an OTSU (maximum between-class variance) image thresholding algorithm, and the false alarm rate of an artificial target detection result is reduced; and the artificial targets are accurately positioned in the target zone, the technical problems of the SAR image target detection with high false alarm rate and low possibility of accurate positioning of the artificial targets in the target zone are solved, and the processing speed is quick. The SAR image target detection method can be applied to the target detection in the field of SAR image processing and computer vision.

The invention discloses a method for quickly implementing ultrasonic imaging on basis of compressive sensing and mainly solves the problem that the time of computing iteration time step length by a quick iterative shrinkage/thresholding algorithm under a compressive sensing frame is slow. The method includes the steps of 1, according to a set resolution, discretizing a detection area; 2, subjecting the discretized detection area to wide-band pulse scanning to obtain an echo vector and an observation matrix; 3, establishing an ultrasonic imaging mathematic model according to the echo vector and the observation matrix; 4, determining the iteration step length by acquiring a 2-norm of a sum of module values of each row of elements in the observation matrix; 5, plugging the iteration step length into the quick iterative shrinkage/thresholding algorithm to solve scattering intensity of a reconstructed observation scenario; and 6, arranging module values of the scattering intensity into a two-dimensional matrix so as to obtain a reconstructed ultrasonic image. Compared with the traditional quick iterative shrinkage/thresholding algorithm, the method has the advantages that the time of computing the iteration step length is greatly shortened, timeliness of the ultrasonic image is improved, and the method is applicable to the field of ultrasonic real-time processing.

The invention relates to a CT (Computerized Tomography) image metal track prediction and artifact reduction method based on integral cosine. Aiming at metal area part cutting in a metal artifact reduction algorithm, compared with a common threshold algorithm which uses original sinogram data to reestablish images, then cuts an image domain into metal pixels and finally projects the cut metal pixels onto a sinogram domain to position a metal track, the CT image metal track prediction and artifact reduction method based on integral cosine firstly obtains various projection tracks according to a cosine projection relationship, projections on each track are added and values which are greater than a threshold are regarded as values corresponding to metal, so that the metal track is obtained. The CT image metal track prediction and artifact reduction method based on integral cosine has the advantages that the original images are not required to be reestablished, the time is greatly saved, the accurate positioning of metal areas in images is kept as much as possible and artifacts introduced in the reestablished images are ensured to be as few as possible. According to results, the CT image metal track prediction and artifact reduction method can be used in a metal artifact reduction algorithm of CT medical images and the metal artifact reduction effect of the repaired images is better.

The invention provides a bottled liquid impurity and liquid level detection method based on image processing. The bottled liquid impurity and liquid level detection method comprises following steps: a, images to be processes are collected and inputted by a camera; b, a memory space used for storing images obtained via each step of treatment is constructed; c, Gaussian filtering is adopted for image grayscale smoothing treatment; d, improved otsu auto-thresholding algorithm is adopted so as to obtain a threshold T1; e, the threshold T1 is subjected to amplification so as to obtain a new threshold T2; f, the threshold T2 is used for binary-morphology treatment of the images obtained via Gaussian filtering; g, detection on impurities and liquid level is carried out, and results are outputted; and h, the memory space is released. The bottled liquid impurity and liquid level detection method is capable of realizing self-adaptive adjustment of thresholds in a certain light range. It is shown by experiment results that the bottled liquid impurity and liquid level detection method is capable of identifying impurities with a diameter larger than 0.5mm, detecting liquid level accurately, giving an alarm on unqualified products, and rejecting unqualified products.

The invention discloses a method and system for extracting information from a received signal with the minimum loss due to noise. The system includes: a converter for correlating the received signal with a wavelet function and generating wavelet decomposition coefficients; and a threshold circuit, responsive to the received signal, for applying a predetermined threshold based on the type of signal. The system also includes a filter connected to the transformer and a threshold circuit for altering the wavelet decomposition coefficients produced by the transformer using a threshold applied by the threshold circuit to produce altered wavelet coefficients from which The reduced noise reconstructs the received signal.

A method for training a neural network to be configured to filter a multipath corrupted signal is provided. The method includes receiving, at the neural network, real or simulated multipath corrupted signal data, and training the neural network on the multipath corrupted signal data using a complex iterated least square thresholding algorithm (CILST) capable of processing both real and complex signals.

The invention discloses a bridge moving load identification method based on cascaded dictionaries and sparse regularization. The method mainly includes the following steps: 1) arranging a speed measuring device on a bridge to be used for measuring vehicle speed, and arranging a plurality of strain and acceleration sensors on the bridge for actual measurement of bridge response; 2) building a bridge finite element model; 3) adopting cascaded dictionaries formed by a discrete trigonometric function and a discrete rectangular function to expand an unknown moving load; 4) normalizing actual measurement response, and introducing a weighted L1 norm regularization method to build an identification equation; and 5) utilizing a fast iterative shrinkage-thresholding algorithm (FISTA) to solve the equation, and calculating a corresponding time interval signal. The method provided by the invention can accurately identify a bridge moving load, including a slowly-changing integral signal and local impact signal in bridge moving vehicle load signals, has strong noise robustness, and is particularly suitable for field utilization of bridge response to estimate gross vehicle weight.

The invention discloses a portrait paper-cut generation method, which comprises the following steps: decomposing a portrait paper-cut according to five sense organs of a human body so as to respectively establish an eyebrow subtemplate library, an eye subtemplate library, a nose subtemplate library, a mouth subtemplate library and an outer contour subtemplate library; receiving an input image to obtain positioning coordinates of eyebrows, eyes, a nose, a mouth and an outer contour of a person in the input image; extracting feature regions of the input image according to the positioning coordinates, wherein each feature region corresponds to an image block; processing each image block by using an OSTU based dynamic threshold algorithm so as to generate a binary sequence of each image block; and computing the distance between each binary image in the binary sequence of the i(th) image block and a corresponding image block so as to find out a window size index number corresponding to a binary image with the minimum distance, and carrying out binarization processing on the input image by using a window size corresponding to the window size index number so as to obtain a binary image set. According to the invention, the artistic characteristics of paper-cuts are added into a portrait paper-cut algorithm, thereby ensuring the artistic properties of portrait paper-cuts.

The invention provides a steel rail surface defect identification and classification method, which is based on unmanned aerial vehicle monitoring image and depth forest model analysis, and comprises the following steps of: firstly, preprocessing an unmanned aerial vehicle monitoring image; secondly, extracting steel rail surface defects based on a proportional reinforcement maximum entropy threshold algorithm; and finally, realizing small sample defect data classification by utilizing a depth forest method. The steel rail surface defect identification and classification method provided by theinvention has the advantages and positive effects that the blind zone problem of a traditional detection mode in the field of steel rail surface defect detection is solved, the improved method for processing the unmanned aerial vehicle image is provided, and a certain promotion effect is achieved on track defect detection by utilizing the unmanned aerial vehicle; and the defect classification method suitable for the small samples is provided, and the problem of insufficient sample data volume at present can be solved on the basis of ensuring the accuracy.

The invention discloses a sign early warning method and system for a participant. The method comprises the steps that a cloud server acquires the sign parameter historical data and the identificationinformation of the participant in advance, wherein the cloud server comprises a sign safety threshold algorithm model; previously acquired sign parameter historical data of mass participants are usedas basic data, the safety threshold of dynamic sign data corresponding to each participant is acquired through the mode of big data operation; a wearable device acquires the real-time dynamic sign data of the participant during a competition; real-time monitoring data including the real-time dynamic sign data are uploaded to the cloud server; and the cloud server generates the sudden death risk assessment result of the participant according to the safety threshold of the dynamic sign data of the participant and the real-time dynamic sign data.

The invention discloses a non-local neural network myocardial transmembrane potential reconstruction method based on an iterative shrinkage threshold algorithm. Through the fusion of a neural networkand a conventional optimization algorithm, and according to the non-local self-similarity of a transmembrane potential in space-time distribution, the invention designs a non-local feature extractionmodule; and an iterative shrinkage threshold algorithm optimization framework provides reliable theoretical explanation and convergence guarantee for the algorithm, the fusion of the neural network increases the flexibility and generalization performance of the method, and the addition of the non-local feature extraction module improves the performance of the network. Through the finally obtainedspace-time distribution of the myocardial transmembrane potential, the position and boundary of an infarction area during myocardial infarction, a pace-making point during ectopic pace-making and other areas with abnormal electrical activity can be accurately observed through three-dimensional display or waveform display, and the method has important reference significance in clinical diagnosis and treatment.

The invention discloses a Kalman filtering steel truss bridge structure parameter monitoring method based on adaptive control. The method comprises the following steps: firstly, determining mechanical parameters such as mass and rigidity of a steel truss bridge structure; secondly, speed, acceleration, displacement and ground acceleration vibration response data of a truss where the steel truss bridge structure is located are collected; and then tracking the change of the speed, acceleration, displacement and rigidity of the structure by using a Kalman filtering algorithm, performing envelope control by combining a dynamic statistical process algorithm, and automatically alarming when the rigidity generates a change value and exceeds an envelope control line. According to the method, the Kalman filtering algorithm is optimized so that the Kalman filtering algorithm can achieve the change of the predicted rigidity on the steel truss bridge structure, and the actual engineering requirements are met; compared with a traditional single threshold algorithm, the defect of a fixed threshold is overcome, the fixed threshold is optimized into dynamic control, reference is provided for performance evaluation of the steel truss bridge structure, and management and maintenance work of the steel truss bridge structure can be effectively served.

The invention discloses a compressed-sensing-based restoration method of a two-dimensional code image with non-uniform illumination. The method comprises the following steps that: a same two-dimensional code image is collected to obtain two two-dimensional code images and sparsity analysis is carried out on the two-dimensional code images; Fourier transform is carried out on the two two-dimensional code images to obtain spectrum images X1 and X2, a non-correlation sampling matrix is set according to the Fourier features of the two-dimensional code images, wherein the non-correlation sampling matrix is formed by a cross and a concentric circle and the sampling cross width and concentric circle dimension can be adjusted properly, and spectrum sampling is carried out on the spectrum images X1 and X2 at the Fourier zone so as to obtain a Y1 and a Y2; linear fusion is carried out on the Y1 and Y2 to obtain a new Fourier spectrum Y; and quick soft threshold iteration is carried out on the Fourier spectrum Y by using a fast iterative shrinkage-thresholding algorithm (FISTA) to obtain a Y', Fourier transform is carried out on the Y' to obtain a restored image f, and binarization processing is carried out on the image f and a two-dimensional code image is identified. With the method, equalized illumination can be realized; a processing speed of an image with non-uniform illumination can be accelerated; and an original two-dimensional code image can be restored accurately.

The invention provides a high-precision settlement monitoring method and device for SAR data, and relates to the technical field of data processing. The method comprises the steps: obtaining to-be-processed SAR data inputted by a user, and enabling the to-be-processed SAR data to be SAR data in a TOPS mode; performing high-precision registration processing on the to-be-processed SAR data to obtaina multi-scene image; determining a pre-selected point of a scatterer of the multi-scene image by utilizing an amplitude deviation threshold algorithm; performing phase correction processing on the pre-selected point to obtain a target preset point, and determining a target scatterer by using the target preset point; and performing phase unwrapping processing and atmosphere and orbit correction processing sequentially on the target scatterer to obtain linear deformation information of the area corresponding to the to-be-processed SAR data. The technical problem that an existing SAR data settlement monitoring method is low in monitoring precision of the linear deformation information of the monitoring area is solved.

The invention discloses a fire point detection method of Himawari-8 remote sensing data, and belongs to the technical field of fire point detection of remote sensing data. The method comprises the following steps: preprocessing Himawari-8 satellite remote sensing data to obtain pixels and neighborhoods thereof, and making a data set; building a double-input convolutional neural network model, and inputting the radiation temperature wave band and the reflectivity wave band into the network; training a deep learning model by using the data set, iteratively updating network model parameters according to the change condition of a loss function, and converging to obtain an optimal network model; and applying the obtained model to fire point detection of data in a test set. Compared with a traditional context relative threshold algorithm, the method provided by the invention can automatically carry out fire point judgment, and can effectively improve the accuracy of fire point detection of remote sensing data.

The invention relates to the field of artificial intelligence, and provides a choroidal vessel lumen area recognition method, device and equipment and a medium. The method comprises the steps of obtaining a fundus image to be recognized; inputting a fundus segmentation model based on U-Net, and performing choroidal feature extraction and edge segmentation on the fundus image to be recognized through the fundus segmentation model to obtain a fundus segmentation image; recognizing a foveal area in the fundus segmentation image through a fundus foveal recognition model, and intercepting a first fundus choroid image from the fundus segmentation image according to the foveal area; carrying out binarization processing on the first fundus choroidal image through a Niback local threshold algorithmto obtain a first choroidal binary image, and extracting a first lumen area from the first choroidal binary image; and identifying the first lumen region image. According to the invention, the funduschoroidal vessel lumen area in the fundus image can be automatically identified. The invention is suitable for the field of smart medical treatment, and can further promote the construction of smartcities.

The invention discloses an intelligent electric meter chip image binarization processing method based on an adaptive mixed threshold. The method comprises the following steps: S1, preprocessing an original image to obtain a grayscale image; s2, calculating a global threshold Tg according to the gray value of the whole gray image; s3, calculating a stroke width corresponding to each pixel point inthe grayscale image, adaptively calculating a window width corresponding to each pixel, and calculating a local threshold Tw of each pixel according to a pixel value in each window; s4, combining theglobal threshold Tg and the local threshold Tw to calculate a mixed threshold T; and S5, performing binarization operation on the grayscale image according to the mixed threshold T to obtain a binarized image. Compared with the prior art, the processing method provided by the invention improves the traditional local threshold method for the problem of non-uniform brightness of the chip image of the intelligent electric meter, and performs weighted addition on the global threshold and the local threshold, thereby solving the problem that the local threshold algorithm does not consider the overall effect of the image.

The invention discloses a CT image-based hepatic blood vessel three-dimensional segmentation method. According to the method, adaptive S-type nonlinear gray mapping and isotropic interpolation sampling are adopted; and improved hepatic blood vessel enhancement processing based on Hessian matrix eigenvalue analysis is carried out; blood vessel enhanced response is introduced into a fuzzy affinity function; fuzzy segmentation parameters are initialized by means of an Otsu multi-threshold algorithm; and a complete hepatic blood vessel is obtained through a fuzzy connection segmentation algorithm.With the method adopted, a complete three-dimensional hepatic blood vessel structure can be segmented from a CT image, and segmentation accuracy is significantly improved.

The invention discloses a wireless sensor node task time slot scheduling method based on matrix completion, and the method comprises the steps: dividing the continuous working time of a node into different task time slots through taking a sampling time slot as a unit based on the time correlation and low-rank characteristics of sampling data of a wireless sensor node in an environment monitoring application; the energy consumption of the node is reduced and the battery life of the node is prolonged by reducing the frequency of executing a sampling task time slot by the node; constructing a sampling matrix of wireless sensor nodes at a coordinator end, reconstructing a node sampling matrix subjected to data reduction by utilizing a singular value threshold algorithm, ensuring that completenode sampling data is provided for a user, and meanwhile, meeting the requirement of low-power-consumption operation of the wireless sensor network nodes; designing a singular value threshold algorithm of dynamic correction amplitude, adjusting the correction amplitude according to the reconstruction recovery precision of the sampling matrix, reducing the iteration frequency of matrix reconstruction, shortening the reconstruction time of the sampling matrix, and improving the algorithm performance.