68 results about "Otsu thresholding" patented technology

Apparatus for use detection of apnea includes a microphone mounted in the ear of the patient for detecting breathing sounds and a second external microphone together with an oximetric sensor. A transmitter at the patient compresses and transmits the signals to a remote location where there is provided a detector module for receiving and analyzing the signals to extract data relating to the breathing. The detector uses the entropy or range of the signal to generate an estimate of air flow while extracting extraneous snoring and heart sounds and to analyze the estimate of air flow using Otsu's threshold to detect periods of apnea and/or hypopnea. A display provides data of the detected apnea/hypopnea episodes and related information for a clinician.

The invention discloses a color image segmentation method, which comprises the following steps: processing an image to be segmented into a gray image; extracting a region profile map from the gray image by an Otsu threshold segmentation method, and determining an optimal threshold; taking the optimal threshold as a high threshold of a Canny operator, and extracting an edge map from the gray image by utilizing the Canny operator; and fusing the area profile map and the edge map, and outputting a segmentation result of a color image. The method adopts an operator threshold determining scheme combined with an adaptive strategy and an empirical value and overcomes the defect that an empirical value selecting method has application limitation; the local change intensity of a gray value is obtained through edge detection, so that over-segmentation of avoidable areas is limited; and residual edges are complemented through region segmentation, so that profiles of images after the segmentation are more clear and complete.

The invention discloses a video based multi-vehicle traffic information detection method. The method includes the steps of acquiring traffic video; setting parameters; initializing a system; detecting a vehicle object; combining an initial color background image with an RGB (red, green, blue) color image in current frame of a video to obtain a self-adapting real-time dynamic color background image; extracting a color difference result image fgi; segmenting a Otsu threshold in a self adaptation manner; removing shadow off a foreground target image; performing morphological operation and filling car mass; counting cars; judging whether the car is detected by at least one virtual detecting coil or not, if yes, determining that the car is detected by the virtual detecting coil, adding 1 to total number of the cars, and executing the step 5; if not, executing the step 406; and acquiring traffic information of the multiple vehicles. By the aid of the video based multi-vehicle traffic information detection method, every passing car is traced, vehicle type and speed of the car are recorded, traffic information of the multiple vehicles such as flow and average speed of different vehicles are obtained, lane crossing vehicles, adhesion and shielding factors of the vehicles are fully considered, anti-interference performance is strong, and detecting accuracy is high.

The present invention provides an automatic cloud detection method and system for a multi-spectral remote sensing satellite image. The method comprises: preparing data; roughly extracting clouds; extracting texture information on an image intensity information channel by combining histogram equalization and bilateral filtering; segmenting a texture information map by using a two-dimensional Otsu threshold; removing an error from a rough detection result by using a binary detail map after segmentation; using intensity information of a raw image as a guide map; and based on the rough detection result with the error being removed, performing accurate extraction on the clouds through edge seed expansion. The technical scheme actually realizes automatic, fast, and accurate detection of cloud.

The invention discloses a lactating sow image segmentation method integrated with FCN and threshold segmentation. The method comprises the steps: collecting a video image of a sow, and building a sow segmentation video image library; building an FCN sow segmentation model, segmenting a test image through the model, and obtaining an FCN sow image segmentation result; building a bounding minimum-area rectangular frame according to the FCN segmentation result, carrying out the Otsu threshold segmentation of the gray scale image and H component of the region, and obtaining a threshold segmentation result; carrying out the fusion of the FCN segmentation result and the threshold segmentation result, and obtaining a final segmentation result of a sow image. On the basis of FCN, the method integrates with the multi-channel Otsu threshold segmentation technology, can effectively complement for a local regional loss while the FCN segmentation effect is not reduced, and improves the segmentation accuracy.

A CIE Lab color space based gray threshold segmentation method comprises the following steps of 1 transforming an image of a RGB color space into a CIE Lab color space, 2 conducting Gaussian histogram filtering on all gray channels of the CIE Lab color space, 3 adopting an Otsu threshold value method to calculate threshold values of the gray channels and adjusting the threshold values as local minimum, 4 calculating gray separation degrees of the gray channels of the CIE Lab color space, selecting the gray channel with highest gray separation degree and adopting the corresponding threshold value calculated in the step 3 to perform binaryzation division. By means of the CIE Lab color space based gray threshold segmentation method, the problem of division of a part of color images is well solved. In addition, the division operation can serve as pre-division operation conducted on complicated color images, and proposed separation degree index can also serve as a standard for image estimation.

The invention discloses a robot fire detection method, comprising the following steps: 1) obtaining a video image of a fire detection area through a high-definition camera, preprocessing the video current frame image in a robot operating system to obtain an image z1; 2) perform flame region segmentation on that image z1 based on an Ohta color space and an Otsu threshold segmentation algorithm to obtain a segmented image f1; 3) subtracting the image z2 after the preprocessing of the previous frame from the image z1 of the current frame by the inter-frame difference method, and segmenting the image into moving regions to obtain a segmented image f2; 4) intersecting the obtained segmented image f1 and the segmented image f2 in the robot operating system to obtain a segmented image f3 having aflame motion characteristic; 5) carry out flame recognition on that region in the segmented image f3 base on other characteristics of the flame; 6) judging whether the flame area is great than a threshold value, and returning the final detection result of the fire. The novel robot fire detection method provided by the invention has the advantages of high accuracy and good real-time performance for identifying multiple features of the flame.

The invention discloses an infrared cloud picture cyclone analysis method, which comprises the following steps of (1) constructing a cyclone system detection network based on an SSD frame of a convolution neural network, and automatically identifying and positioning the cyclone from each cloud picture in a cloud picture data set by the network; 2) carrying out the threshold segmentation on the cyclone by adopting an OTSU threshold segment algorithm, and carrying out the area filtering on the threshold segmentation result to obtain the initial contour of the cyclone; (3) taking the initial contour of the cyclone as input, using a Chan-Vese model to obtain a cyclone boundary of the cyclone; (4) extracting cyclone characteristic points of the cyclone boundary by using the SURF algorithm; (5)using an FLANN matcher to calculate the matching ratio of each characteristic point of cyclone at adjacent time by Euclidean distance, so that the cyclone can be tracked. The invention draws lessons from the MobileNet network and combines the SSD detection frame to construct the cyclone detection network for recognizing the cloud picture, which has high recognition accuracy and is simpler.

The invention relates to a method for a service robot to recognize and locate a target, and the method comprises the following steps of (1) utilizing two identical charge coupled device (CCD) cameras to simultaneously observe on target point so as to acquire a perceived image of the target point under different viewing angles, and calculating a position deviation of every two adjacent image pixels through a geometrical imaging principle to obtain three-dimensional information of the target point; (2) utilizing an image processor to process the perceived image, and utilizing a conversion formula to transform red-green-blue (RGB) color space to hospital information system (HIS) color space; (3) acquiring an image segmenting threshold value through an Otsu threshold value method; and (4) utilizing the image segmenting threshold value to segment the perceived image into correspondent communication areas, utilizing a color characteristic to mark the area which is interested by the image, and completing the recognition and location of the target. Compared with the prior art, the method has advantages that influences of factors such as complexity and nonuniformity of the lighting condition can be reduced, the validity and accuraty for segmenting the image can be greatly improved, and the like.

The invention discloses a three-dimensional reconstruction method for a defective blade tip of an aircraft engine compressor blade. The method is characterized by comprising the following steps that 1, a repairable worn blade is measured through an optical three-dimensional measurement method to obtain point clouds of the blade; 2, error point clouds generated in three-dimensional measurement are removed through a smooth filtering method, and the point clouds are simplified through a bounding box method; 3, a maximum value of between-class variance between a boundary point cloud and a curved surface point cloud is calculated by using an Otsu threshold value method for reference, the boundary point cloud and the blade point clouds are distinguished, whether the boundary point cloud is on the boundary of a defectless area or the boundary of a defective area is judged by comparing specific values of principle curvatures, in the two mutually perpendicular directions, of the boundary point cloud, and the obtained boundary point cloud is fit into a boundary line through a least square method; 4, the point clouds of the portions except for the defective portion are fit into a B-Spline curved surface; 5, similarity extension is conducted on the boundary point cloud of the defective area to obtain a complete three-dimensional model. After detection, the similarity measuring value of extension is not larger than 0.2 mm.

The invention discloses an improved convolutional neural network-based plant leaf segmentation method. Eight-time up-sampling of original FCN is large in model parameter amount; in order to realize rapid and accurate segmentation of leaves, a direct connected structure is adopted to remove a part of the layers, single layer features are utilized to up-sample feature maps shrunk by a VGG16 model toan original map size according to a shrinking multiple through deconvolution, and three improved models are obtained. At last, 1762 pictures of 6 different plants are taken as training samples, 441 leave pictures are taken as test sample training models. Memory parameter occupation of direct connection type structure models which adopt four-time up-sampling is 6.90 MB, the shrinkage is 1/77 timeof the original FCN model, and the average correctness and average area overlap ratio of categories on test sets can be up to 99.099% and 98.204%. Compared with the tradition K-means cluster method and Otsu threshold value segmentation method, the method has the advantages that the average area overlap ratio is 3.704% and 4.295% higher, complete leaves can be extracted, and the influences of leavesurface color and illumination intensity unevenness are small.

The invention discloses a method for detecting remote sensing image change based on adaptive Treelet construction difference images, belongs to the field of image processing technology and mainly assists in solving a problem of change detection precision deficiency in the prior art. The method is realized by carrying out block processing of two inputted different-phase remote sensing images, namely calculating difference values between an image block in a search window of a second image and a center image block of a first image to obtain sample matrixes; clustering the sample matrixes by using Treelet algorithm to obtain adaptive difference images; calculating Otsu thresholds of differential-value difference images and adaptive difference images; integrating the differential-value difference images and the adaptive difference images by using the thresholds to obtain final difference images, and carrying out Otsu threshold segmentation on the final difference images to obtain change detection results. According to the method in the invention, change detection precision can be improved effectively; image edge information can be maintained; and the method can be used for detecting disaster situation and in land utilization.

The present invention discloses a device and method of on-line automatic count of fries on the basis of machine vision. The device provided by the invention comprises a red LED backlight, an industrial camera, pipelines, a filter screen, a round plane, an uncovered cylindrical sink, electromagnetic valves, a liquid level meter, a PC machine containing an image acquisition card, a single-chip microcomputer controller and a support. The acquired images are subjected to arithmetic processing such as grey level transformation, Laplace filtering, adaptive median filtering, OTSU threshold segmentation, morphological processing, endpoint refinement and the like and then fry counting through necessary software. The device provided by the invention is able to be continuously and effectively used for a long time under the control of the PC machine, so that deficiencies of traditional manual counting, such as limited working time, low precision, obvious subjective factors and the like, are solved; and the water level may be detected and controlled through the liquid level meter to adapt counting of different fries with different sizes so that different users' counting requirement may be satisfied.

The invention provides a tobacco field image segmentation method. The method is a farmland image segmentation method based on Lab and YUV color space. The method is characterized by merging a binary image obtained through Otsu threshold segmentation based on the Lab color space and a binary image obtained through a weighting fuzzy entropy segmentation method based on the YUV color space, and then, carrying out filtering so as to obtain an optimum segmentation effect. According to the experiment, the method can filter the noise and suppress the influence caused by complex environment of non-uniform illumination and the like better, and thus a satisfactory segmentation result is obtained, and accurate segmentation of the farmland image is realized.

The invention discloses a video watermarking method based on a depth image and Otsu segmentation. The method comprises the steps of obtaining the depth image of a video keyframe, scrambling a watermark picture through Logistic mapping to generate disordered one-dimensional watermark information, segmenting the depth image into a foreground region and a background region through an Otsu threshold segmentation method according to depth-of-field information provided by the depth image, judging the foreground region of the video keyframe, embedding the watermark information in a DCT coefficient of luminance component subblocks belonging to the foreground region, judging the foreground region and the background region of a video containing a watermark and extracting the watermark information from the DCT coefficient of the luminance component subblocks belonging to the foreground region. According to the video watermarking method, the embedded region of the watermark is determined according to the depth information of the video keyframe, so the problem that scene spatial position relativity is not considered in a human visual system is effectively solved, and good robustness to attacks of pepper and salt noise, multiplicative noise, gaussian noise, luminance contrast adjustment and the like is achieved.

The invention discloses a medical image segmentation method and a device, wherein the method comprises: step S11) obtaining magnetic resonance angiography image; step S12) for the magnetic resonance angiography image, using the Otsu threshold method to divide it into an interested foreground area and a background area; and calculating difference value between the pixel mean gray value of the foreground area and the pixel mean gray value of the background area corresponding to the maximum of the variance function of the foreground area and the background area; and Step S13): determining the difference value between the gray mean values of the internal images and external images of the evolution curve of the image segmentation model C-V according to the difference value; and segmenting the magnetic resonance angiography image according to the determined difference value between the gray mean values of the internal images and external images of the evolution curve and obtaining a segmenting result. The medical image segmentation device includes an image obtaining module, a difference value determining module, and a segmenting module. The medical image segmentation method and the device of the present disclosure improve the segmenting effect and the processing speed, meeting the requirements.

The embodiment of the invention provides a method and a device for identifying and classifying cotton pests. The method obtains an original image containing cotton pests to be classified, and obtainsHu invariant moment parameters of the original image and overall contour characteristic parameters of cotton pests to be classified. Based on Otsu threshold segmentation algorithm and Canny edge detection algorithm, the image of cotton pests to be classified is separated from the background of the original image, the wing image is extracted from the original image, and the wing contour characteristic parameters of cotton pests to be classified are obtained according to the wing image. Mathematical morphology algorithm is used to optimize the wing image and extract the corresponding mathematical morphological parameters of the wings of cotton pests to be classified. Radial basis function neural network is used to classify cotton pests. The method and the device enable the identification andclassification of cotton pests to be more accurate, thereby playing an important role in the targeted control of cotton pests and reducing the economic losses caused by cotton pests.

The invention provides a robust global threshold segmentation method. The method includes the steps that step1, a threshold value To of a gray-level image is calculated through an Otsu threshold value method; step2, a balance degree factor eta is defined through the variance of a foreground and a background of the gray-level image, and a threshold value Tb with the balance degree factor as the criterion is acquired; step3, a deviation degree lambda and a new optimal criterion zeta are defined according to Tb and To, and a segmentation result is generated. By means of the method, the optimal threshold value can be obtained under the condition that the probability distribution difference between the foreground and the background of the grey-level image is obvious or under general conditions.

The invention discloses a method for removing shadows and halos under weak light through an indoor polling robot. The method comprises the following steps: carrying out multiscale Retinex strengthening on a picture collected by the robot; carrying out limited times of iteration on pixel points of the strengthened image through mean shift filtration; eliminating the noise points; smoothing the image; removing the low-light shadow areas by utilizing an illumination normalizing method; and carrying out image segmentation through an OTSU threshold segmentation method so as to obtain a binary result. According to the method, the shadows of equipment in weak light environment can be effectively removed, the influences of the halos can be eliminated, and the effective information required by the later-period mode recognition can be kept.

The invention discloses an improved robust two-dimensional OTSU threshold image segmentation method. By using median filtering on an original image in an original two-dimensional OTSU algorithm, the robustness of the algorithm to noises such as salt and pepper is enhanced; Partitioning the to-be-processed image, and redefining inter-class variance measurement of a two-dimensional OTSU algorithm based on a plurality of regions to realize threshold segmentation of the image with the non-uniform background gray scale; The efficiency of the algorithm is improved by replacing a threshold exhaustionstrategy in a two-dimensional OTSU algorithm with univariate iteration. The method has a good segmentation effect on a noise-added image and an image with non-uniform background brightness, and the threshold search algorithm can reduce the time cost of the search process.

The invention discloses a high-altitude object throwing detection method based on computer vision and radio signal analysis. The detection method comprises a computer-vision-based method and a radio-signal-analysis-based method. The computer-vision-based method includes a background-modeling-based mobile detection method, a background-modeling-based mobile detection algorithm, an adaptive OTSU threshold calculation method, and a self-made high-speed tracking algorithm. According to the invention, the high-altitude object throwing time and location can be captured by means of computer vision and radio analysis. On the basis of combination of real-time analysis with video recording, the high-altitude object throwing time and location can be located accurately and thus evidences are taken accurately; and the capturing delay of the camera is avoided. Meanwhile, the high-altitude object throwing can be detected precisely by using the computer vision and radio signal analysis; and the false reporting number is reduced. Besides, a monitoring camera and a lens are arranged to monitor the high-altitude object throwing, so that objects thrown at high altitudes can be prevented from being moved or picked up and thus the evidences are protected from being damaged.