Patents
Literature
Eureka-AI is an intelligent assistant for R&D personnel, combined with Patent DNA, to facilitate innovative research.
Eureka AI

473 results about "Gradient direction" patented technology

The direction of the gradient is simply the arctangent of the y-gradient divided by the x-gradient. tan−1(sobely/sobelx). Each pixel of the resulting image contains a value for the angle of the gradient away from horizontal in units of radians, covering a range of −π/2 to π/2.

Real-time robust far infrared vehicle-mounted pedestrian detection method

The invention discloses a real-time robust far infrared vehicle-mounted pedestrian detection method. The method comprises the steps of catching a potential pedestrian pre-selection area in an input image through a pixel gradient vertical projection, searching an interest area in the pedestrian pre-selection area through a local threshold method and morphological post-processing techniques, extracting a multi-stage entropy weighing gradient direction histogram for feature description of the interest area, inputting the histogram to a support vector machine pedestrian classifier for online judgment of the interest area, achieving pedestrian detection through multi-frame verification and screening of judgment results of the pedestrian classifier, dividing training sample space according to sample height distribution, building a classification frame of a three-branch structure, and collecting difficult samples and a training pedestrian classifier in an iteration mode with combination of a bootstrap method and an advanced termination method. According to the real-time robust far infrared vehicle-mounted pedestrian detection method, not only is accuracy of pedestrian detection improved, but also a false alarm rate is reduced, input image processing speed and generalization capacity of the classifier are improved, and provided is an effective night vehicle-mounted pedestrian-assisted early warning method.
Owner:SOUTH CHINA UNIV OF TECH

Random convolutional neural network-based high-resolution image scene classification method

The invention discloses a random convolutional neural network-based high-resolution image scene classification method. The method comprises the steps of performing data mean removal, and obtaining a to-be-classified image set and a training image set; randomly initializing a parameter library of model sharing; calculating negative gradient directions of the to-be-classified image set and the training image set; training a basic convolutional neural network model, and training a weight of the basic convolutional neural network model; predicting an updating function, and obtaining an addition model; and when an iteration reaches a maximum training frequency, identifying the to-be-classified image set by utilizing the addition model. According to the method, features are hierarchically learned by using a deep convolutional network, and model aggregation learning is carried out by utilizing a gradient upgrading method, so that the problem that a single model easily falls into a local optimal solution is solved and the network generalization capability is improved; and in a model training process, a random parameter sharing mechanism is added, so that the model training efficiency is improved, the features can be hierarchically learned with reasonable time cost, and the learned features have better robustness in scene identification.
Owner:WUHAN UNIV

Fault identification method of high voltage transmission line based on computer vision

The invention relates to a high-voltage transmission line fault identification method based on computer vision, which relates to the technical field of high-voltage transmission line running state monitoring. The invention aims at solving the problem of high false alarm rate of the existing high-voltage transmission line on-line monitoring system. 11) carrying out edge detection on the transmission line image according to the edge detection algorithm, a strong edge image is obtained, and edge endpoints and edge directions are obtained from the strong edge image. Since the gradient direction ofthe edge endpoints is perpendicular to the edge direction, an edge connection window is selected according to the gradient direction of the edge endpoints, and edge connection points are selected inthe edge connection window according to a Hough transform method, and the edge connection points are connected into an edge image. Step 2, screening the transmission lines from the edge images of thetransmission line images by adopting a transmission line detection algorithm based on phase grouping; Step 3, the transmission conductor is processed to identify the fault on the transmission line. Itis used to identify transmission line faults.
Owner:国网黑龙江省电力有限公司佳木斯供电公司 +2

Rapidly converged scene-based non-uniformity correction method

InactiveCN102538973APrevent erroneous updatesBug update avoidanceRadiation pyrometryPhase correlationSteep descent
The invention discloses a rapidly converged scene-based non-uniformity correction method, wherein the aim of non-uniformity correction is achieved by minimizing interframe registration error of two adjacent images. The method mainly comprises the following steps of: initializing gain and offset correction parameters and acquiring an uncorrected original image; acquiring a new uncorrected original image, and carrying out non-uniformity correction on the new uncorrected original image and the previous uncorrected original image by utilizing the current non-uniformity correction parameters; obtaining relative displacement, scene correlation coefficient and interframe registration error of two corrected images by utilizing an original point masking phase correlation method; and updating correction parameters along the negative gradient direction by adopting a steepest descent method. The method disclosed by the invention has the advantages of high correction accuracy, fast convergence speed, no ghost effect and low calculated amount and storage content and is especially applicable to being integrated into an infrared focal plane imaging system, and the effect of improving imaging quality, environmental suitability and time stability of an infrared focal plane array is achieved.
Owner:NANJING UNIV OF SCI & TECH

Rapid sub pixel edge detection and locating method based on machine vision

The invention discloses a rapid sub pixel edge detection and locating method based on machine vision. The method includes the following steps that firstly, a detection image is acquired; secondly, denoising pretreatment is conducted on the image; thirdly, the gradient Gx of each pixel point in the horizontal direction and the gradient Gy of each pixel point in the vertical direction are calculated; fourthly, the gradient magnitude G0 and the gradient direction Gtheta of each pixel points under polar coordinates are calculated; fifthly, neighborhood pixel points of each pixel point are determined; sixthly, pixel-level edge points are determined; seventhly, the distance between a sub pixel edge point of each pixel-level edge point in the eight-gradient direction and the pixel-level edge point is calculated; eighthly, the distance d between each sub pixel edge point in the actual gradient direction Gtheta and the corresponding pixel-level edge point is calculated; ninthly, a cosine lookup table method is adopted for calculating rectangular coordinates of each sub pixel edge point in the actual gradient direction Gtheta, so that the image edge points are detected and sub-pixel-level localization is conducted. The whole method is high in calculation accuracy and speed.
Owner:湖南湘江时代机器人研究院有限公司

Margin-oriented self-adaptive image interpolation method and VLSI implementation device thereof

ActiveCN103500435AImage enhancementImage analysisVlsi implementationsSynchronous control
The invention discloses a margin-oriented self-adaptive image interpolation method and a VLSI implementation device thereof. The method comprises the steps that the gradient magnitude and the gradient direction of a source image pixel are computed, and marginal information is obtained by comparing the gradient magnitude and a local self-adaptive threshold value, wherein the marginal direction is perpendicular to the gradient direction; the marginal direction is classified, filtering is conducted through the marginal information, and an image is divided into a regular marginal area and a non-marginal area; the regular marginal area interpolation is conducted in the marginal direction, and an improved bicubic interpolation method, a slant bicubic interpolation method and a slant bilinear interpolation method based on local gradient information are adopted to conduct image interpolation according to the classification of the marginal information; image interpolation is conducted on the non-marginal area through the improved bicubic interpolation method based on the local gradient information. The VLSI implementation device comprises a marginal information extraction module, a self-adaptive interpolation module, an input line field synchronous control module and an after-scaling line field synchronous control module. The margin-oriented self-adaptive image interpolation method and the VLSI implementation device of the margin-oriented self-adaptive image interpolation method can effectively improve the effect of image interpolation with high-magnification scaling, and is beneficial to integrated circuit framework achieving.
Owner:XI AN JIAOTONG UNIV

Target tracking method based on multi-characteristic adaptive fusion and kernelized correlation filtering technology

The invention provides a target tracking method based on multi-characteristic adaptive fusion and kernelized correlation filtering technology. The method comprises steps of according to target position and the dimension of the previous frame tracking, acquiring a candidate region of target motion; extracting histogram characteristics and color characteristics in the gradient direction of the candidate region, fusing the two kinds of characteristics, carrying out Fourier transform so as to obtain a characteristic spectrum and then calculating kernelized correlation; determining the position and the dimension of the target at the current frame, and acquiring a target region; extracting histogram characteristics and color characteristics in the gradient direction of the target region, fusing the two kinds of characteristics, carrying out Fourier transform so as to obtain a characteristic spectrum and then calculating kernelized self-correlation; designing the adaptive target correlation and training a position filter model and a dimension filter model; and using a linear interpolation method to update the characteristic spectrums and the related filters. According to the invention, the discrimination capability of the models is improved; robustness of the target tracking of the target in a complex scene and the appearance change is improved; calculation complexity is reduced; and tracking timeliness is improved.
Owner:NANJING UNIV OF SCI & TECH

Water surface optical visual image target area detection method based on gradient information fusion

The invention provides a water surface optical visual image target area detection method based on gradient information fusion. The water surface optical visual image target area detection method based on gradient information fusion adopts two sliding window modes and calculates the longitudinal gradient and the transverse gradient of a water surface optical visual image. The information of the two gradients is fused, the position of an area is marked through a connected component detection method, and the target area is marked according to the final boundary of a target. The water surface optical visual image target area detection method based on gradient information fusion combines sea boundary area characteristics of the water surface image, respectively extracts target boundary information in the longitudinal gradient and the transverse gradient directions of the water surface optical image, determines the attribute of the sea boundary line and division of an image processing space, determines the area type attribute of the target boundary according to the fused information, and finishes scanning and classification of pixels according to the property of the boundary. As the water surface optical visual image target area detection method based on gradient information fusion fuses the information in the two gradient directions by combining the characteristics of the sea boundary lines, reduces the range of the processing area, reduces influence from noise, avoids calculation and processing of the whole image space, and saves calculation time.
Owner:HARBIN ENG UNIV

A scale self-adaptive target tracking algorithm based on kernel correlation filtering

The invention discloses a scale self-adaptive target tracking algorithm based on kernel correlation filtering, and belongs to the field of computer vision. The method comprises the steps of selectinga first frame of to-be-tracked target, and determining as a candidate region; Extracting gradient direction histogram (HOG) characteristics of the candidate area, and carrying out cosine weighting; Performing cyclic shift on the candidate area by using a KCF algorithm to generate a training sample, calculating a response value in a frequency domain according to a position filter obtained by sampletraining and HOG characteristics extracted by the candidate sample, and updating the target position of the current frame; Taking the target position determined by the previous frame as a center to obtain a scale filter, calculating to obtain a response value, and taking the scale corresponding to the maximum response value as the target scale of the current frame; Re-extracting the sample training filter, updating the position filter and the scale filter in a linear interpolation mode, and tracking a subsequent frame; The method can be applied to the fields of intelligent video monitoring, enterprise production automation, intelligent robots and the like.
Owner:NANJING INST OF TECH

Traffic signal lamp automatic detection and recognition method based on visual sense

Provided is a traffic signal lamp automatic detection and recognition method based on the visual sense. A hierarchy framework combining color space linear filtering and gradient direction histogram features is adopted, and a detection and recognition model of typical traffic signal lamps in China is established through training. The method comprises a training stage and a testing stage, in the training stage, collection of a training dataset, determination of a target color spectrum and determination of classification model parameters are completed, in the testing stage, detection and recognition of the traffic signal lamps are achieved by utilizing the models obtained in the training stage, and the testing stage comprises screening of candidate areas of the traffic signal lamps, filtering of connected domains, determining of shapes of the traffic signal lamps and determining of indication directions of the traffic signal lamps. The traffic signal lamp automatic detection and recognition method based on the visual sense overcomes the defects that detection and recognition instantaneity of the traffic signal lamps is poor and the recognition precision is sensitive to illumination and templates in the prior art, achieves better detection and recognition effects, and has wide application prospects in the field of intelligent decision and driver assistant systems of unmanned vehicles.
Owner:BEIHANG UNIV

Real-time-robust pedestrian detection method aiming at specific scene

The invention discloses a real-time-robust pedestrian detection method aiming at a specific scene, comprising the following steps: using a camera to carry out video collection in the specific scene; preprocessing the collected video information at real time by adopting a background difference method to obtain the video information of movable objects in the processed video; taking each movable object in the preprocessed video information as center to demarcate a sliding window; and detecting the image information in the sliding window by a support vector machine pedestrian classifier to match a pedestrian. The feature modeling training of the support vector machine pedestrian classifier comprises the following steps: 1) feature extraction, in which a head and shoulder training sample library is formed by the histogram features at head and shoulder gradient direction and local binaryzation model features; and 2) sample training, in which the head and shoulder training sample library formed by the histogram features at head and shoulder gradient direction and local binaryzation model features is placed into the support vector machine for training so as to obtain the support vector machine pedestrian classifier. The pedestrian detection method has the characteristics of high accuracy and instantaneity.
Owner:BEIJING UNIV OF POSTS & TELECOMM

Bottleneck defect detection method based on gradient direction histograms

The invention discloses a bottleneck defect detection method based on gradient direction histograms and belongs to the technical field of machine vision and image processing. The method comprises steps of: stretching a bottleneck annular part into a rectangle, then performing cutting and splicing, using the modified rectangle as a sample image, subjecting the sample image to Gamma correction to normalize the sample image, dividing the sample image into a plurality of windows, calculating the gradient direction histograms in the windows to obtain a feature vector, forming a classifier by using a support vector machine; then, for a bottleneck image to be detected, obtaining the feature vectors of respective detection windows by the gradient direction histograms, and determining whether a current bottleneck is a defect bottleneck in combination with the preformed classifier. The method, on one hand, extracts pixel gradient information in the bottleneck image based on the principle of the gradient direction histograms so as to realize the real-time detection of the bottleneck defect, and on the other hand, improves the action scope of the defective pixels by an interpolation and normalization method so as to achieve the accurate positioning of the bottleneck defect.
Owner:南京汇川图像视觉技术有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products