184 results about "Adaboost classifier" patented technology

The embodiment of the invention discloses a face recognition method and a device. The method comprises the steps of extracting the Haar feature of a current to-be-recognized face image, and detecting the human face area of the to-be-recognized face image by adopting an ADaBoost classifier so as to obtain a face region image; performing the multi-scale feature extraction on the face region image by utilizing a convolution neural network model, and obtaining a feature vector of the face region image; inputting the feature vector, a pre-built legal face database and a preset user similarity threshold value into a multi-task learning model pre-constructed according to a Softmax loss function and a Triplet loss function, and judging whether the to-be-recognized face image is a legal user or not according to the output value of the multi-task learning model. The extracted feature is good in robustness and good in generalization ability. Therefore, not only the face recognition rate improved, but also the accuracy of face recognition is improved. The safety of identity authentication is improved.

The invention discloses a fast 3D face identifying method based on double-eye passive solid sight, which includes the following steps: 1) a non-contact short shaft parallel binocular stereo vision system is built by applying two high-definition digital cameras; 2) after system calibration is finished, face detection and collection based on a haar-AdaBoost sorting machine is carried out on a preview frame image for obtaining corresponding upper and lower stereoscopic vision graph pairs and estimating a sight difference; image correction is carried out on a face area for obtaining the upper and lower stereoscopic vision graph pairs vertical to the polar lines inside and outside the area; 3) the accurate location on the eyes and a spex nasi is captured by applying a Bayesian and the haar-AdaBoost sorting machines as well as point cloud 3D information for building a benchmark triangle; 4) the corresponding sub pixels in the middle and small areas are matched by applying the pyramidal parallel search solid graph of a phase relevant arithmetic based on a complex wavelet; 5) pose normalizing and hole filling are carried out on the faces under different poses by applying the built benchmark triangle; 6) expression normalization is carried out on different faces based on the suppose that the surface geodesic distance of the face is invariable; 7) the 3D faces after normalization are identified by utilizing the arithmetic. The method has the beneficial effects of: mainly solving the problems of being hard to fast and automatically obtain the passive stereoscopic vision and identifying the 3D point cloud information of the dense and accurate face under different poses and expressions, thus leading the 3D face identifying process to be faster, more hidden, safer and more reliable.

The invention relates to a multi-level-point-set characteristic extraction method applicable to ground laser radar point cloud classification. Based on point set characteristics, high-precision classification of four kinds of common ground features including pedestrians, trees, buildings and automobiles and the like in a scene is realized. Firstly, point sets are constructed and a point cloud is re-sampled into a point cloud of different scales and thus point sets which are different in size and provided with layered structures are formed through clustering and characteristics of each point in the point sets are obtained; next, an LDA (Latent Dirichlet Allocation ) method is adopted to synthesizing point-based characteristics of all points in each point set into shape characteristics of the point sets; and at last, based on the shape characteristics of the point set, an Adaboost classifier is adopted to train the point sets of different levels so as to obtain a classification result of the whole point cloud. The multi-level-point-set characteristic extraction method has a higher classification precision and has a classification precision, which is far higher than that of point-based characteristics, Bag-of-Word-based characteristics and characteristics based on probabilistic latent semantic analysis (PLSA), in aspect of pedestrians and vehicles.

The invention belongs to the field of machine vision and intelligent control for achieving fast self-adaptation of traffic video monitoring target detection. The method comprises the steps of building an initial training sample bank; training an AdaBoost classifier based on Haar characteristics and a support vector machine (SVM) classifier based on histograms of oriented gradients (HOG) characteristics respectively; and detecting monitoring images frame by frame by employing of the two classifiers, wherein the detection process is divided into steps of predicting of sub-images in a detection frame through the two classifiers respectively, performing of confidence-degree determination on predicted results, adding of prediction labels corresponding to large confidence-degree and the sub-images into additional training sample banks of the classifiers corresponding to small confidence-degree till the size of the detection frame reaches half the sizes of detected images, retraining of the two classifiers by using the updated training sample banks and detecting of a next frame of image till all images are detected, and the final classifiers can be used for detecting targets of vehicles, pedestrians and the like in actual traffic scenes.

The invention discloses a method for performing multi-visual-angle face detection by means of integral channel features. The method is mainly characterized in that three LUV color channels in ten ACF channels are improved for obtaining a gray scale single channel, thereby forming an eight-channel characteristic and realizing quicker feature extraction; four-stage Adaboost cascaded classifier training is performed on the extracted features, thereby forming a cascaded strong classifier which comprises 4096 weak classifiers; and image detection is performed by means of the cascaded classifier and a quick feature pyramid method for quickly and accurately detecting faces. According to the method of the invention, detecting blocks are acquired by means of successive sliding of a sliding window on a characteristic pyramid according to a step length; the detecting blocks are classified by means of the trained Adaboost classifier; overlapped window elimination is performed on the detecting blocks which comprise the faces through a non-maximum suppression method; a final face detection window is kept and detection precision is improved.

The invention provides a pedestrian detection method based on saliency information. The method comprises an offline training step and an online detection step; the online detection step comprises calculating a salient map of an image to be detected, extracting a detection child window from the image and calculating the corresponding saliency of the detection child window according to the salient map, calculating corresponding features in the detection child window, detecting the corresponding features in the detection child window through a cascade classifier, and simultaneously distributing adjustment coefficients for the cascade classifier according to the corresponding saliency of the detection child window. According to the method, the saliency information is introduced to be served as auxiliary information for pedestrian detection to participate in the process of image identification on the basis of the existing AdaBoost classifier. In most cases, pedestrians are different from the surrounding environment in terms of color, shape and profile, the saliency information of the child window is adopted for correcting detection results of the classifier, the detection rate can be effectively improved, and the false detecting rate can be reduced.

The invention discloses a pedestrian detection model training method based on an AdaBoost classifier. The pedestrian detection model training method comprises the steps of firstly, conducting real-time statistics on the sum of sample weight values in the AdaBoost training process, when degeneration is carried out to a certain extent, using a currently-trained weak classifier set for scanning a non-pedestrian image for a false detection window, using the false detection window as a difficult sample to be added in negative sample training sets, and decreasing a degeneration degree threshold value so as to reduce sample update efficiency; finally, removing a part of negative samples through random sampling, and reducing the number of the negative sample training sets so as to reduce the calculated amount of the training process. According to the pedestrian detection model training method based on the AdaBoost classifier, on the premise that a feature extraction method is not changed, the training effect of the classifier can be improved to the maximum extent, and the final detection precision is improved.

The invention discloses a detection method and a detection system for video image text and relates to the technical field of image text detection. The method comprises the following steps of coarse positioning of video image text areas and verification of video image text areas. The specific steps are as follows: (A) coarse positioning of video image text area: the gaussian pyramid multi-scale image of a video image is calculated, the image text area is segmented by threshold segmentation, text areas are combined by the inflation and corrosion technology in mathematical morphology, and candidate text areas are analyzed and positioned by connected domain analysis, text line segmentation and elimination rules; and (B) verification of video image text areas: an adaboost classifier is trained by adopting harr characteristics, and the adaboost classifier is utilized to detect whether text exists in the candidate text areas and the candidate text areas the text of which is not detected are eliminated. The method and the system can improve the accuracy and the recall ratio of video image text detection.

The invention relates to the technical field of computer vision, in particular to a multi-feature and multi-threading security inspection contraband automatic identification method based on machine learning. The multi-feature and multi-threading security inspection contraband automatic identification method comprises the steps of: step A, training an Adaboost classifier, wherein target samples are collected, the target samples are preprocessed and subjected to detection and classified identification after feature extraction, thus a strong classifier is obtained; and step B, automatically identifying security inspection contraband, wherein X-ray images of security inspection monitoring are read in, the X-ray images are subjected to PDE denoising pretreatment, parallel edge textures are scanned, parallel feature extraction and classified identification are carried out by utilizing LBP+HOG features and the strong classifier obtained in the step A, and an identification result is output. The multi-feature and multi-threading security inspection contraband automatic identification method solves the problems that phenomena such as wrong inspection and missing inspection are caused by carelessness because the security inspection personnel is tired after monitoring a screen for a long time, the security inspection personnel needs to be on duty for a long time, manpower is wasted and the efficiency is low, greatly improves the working efficiency and reduces manpower cost.

The invention discloses a method for detecting and tracking a vehicle based on the machine vision technology. The method comprises the step of detecting a structuralized road, wherein a picture is preprocessed, the edges of the picture are enhanced and are mediated in a linear mode, the edges of the road are extracted, and the road region is determined; the step of detecting the moving vehicle, wherein the vehicle in the determined road region is initially recognized, the region of interest is further diminished, and whether a vehicle to be detected exists in the determined road region or not is recognized for the second time; the step of tracking the moving vehicle, wherein the accurate position of the target vehicle is determined through repeated iteration tracking based on the mean value drifting algorithm. The invention further discloses a system for detecting and tracking the vehicle based on the machine vision technology. The vehicle can be distinguished more effectively based on the maximum variance threshold value extraction method, the vehicle can be detected more accurately based on the fusion character algorithm, the region of interest is determined, the secondary shrunk ROI region is detected through an adaboost classifier, the operand is greatly reduced, the real-time performance is improved, and meanwhile the accuracy that the system detects the vehicle is improved.

The invention discloses a retinal vessel segmentation method of a fundus image based on a classification and regression tree and AdaBoost. The method comprises the step of: constructing a 36-dimensinal feature vector including a local feature, a morphological feature and a pixel vector field divergence feature for each pixel point in the fundus image, so as to determine whether the pixel point is on a vessel. During classified calculation, the classification and regression tree is used as a weak classifier, so as to classify a sample set, then an AdaBoost classifier is trained, so as to obtain a strong classifier, and thus, the classified determination of each pixel point is completed, so as to obtain final segmentation results. The method has the advantages that a vessel trunk is preferably extracted, great advantages are taken to treat high-brightness focal areas, later treatment is facilitated and visual results are provided for main vessel diseases, and the method is suitable for computer aided quantitative analysis of the fundus image and disease diagnosis and has obvious clinical significance in auxiliary diagnosis of related diseases.

The invention relates to a distracted driving behavior early warning method based on a cab near-infrared camera. The method comprises the following steps of (1) face detection of a driver: acquiring the near-infrared image data of a cab scene offline through the vehicle-mounted near-infrared camera, marking the image data face area, and performing the offline training on an Adaboost classifier serving as a face detection classifier, and searching the face interested region by the face detection classifier; (2) dangerous driving behavior detection: based on the face interested region, adding abackground region, and performing the dangerous driving behavior detection by utilizing a deep convolution neural network; (3) dangerous driving behavior early warning: performing the multi-frame confirmation on the dangerous driving behaviors by means of the time sequence data, if the set continuous frame number threshold value reaches is reached, sending a visual early warning signal and/or an auditory early warning signal. By means of the method, the mental dangerous driving behaviors of the driver can be recognized and warned, so that a driver can be reminded to change the distraction behavior immediately, and the driver can keep the driving state in a normal standard, and the driving safety is improved.

The invention provides a measurement method for a fetus ultrasound image. The measurement method includes acquiring a fetus ultrasound image; establishing an Adaboost classifier of a head circumference or abdominal circumference of the fetus; performing rectangle scanning to the head circumference or abdominal circumference of the fetus according to a sliding-window detecting method, and acquiring a rectangle subimage of the head circumference or abdominal circumference; classifying the rectangle subimage based on the Adaboost classifier, and acquiring interested area of the head circumference or abdominal circumference; extracting edge features of the head circumference or abdominal circumference of the interested area; performing image match to the edge features, and acquiring an ellipse of the head circumference or abdominal circumference; and calculating the length of the ellipse of the head circumference or abdominal circumference. In the process of performing the rectangle scanning to the head circumference or abdominal circumference by adopting the sliding-window detecting method to acquire the rectangle subimage of the head circumference or abdominal circumference, Prior knowledge of the fetus and scanning depth and pixel number information during the clinical test are added, so that detecting speed and accuracy of the interested area of the head circumference or abdominal circumference are improved.

The invention relates to a brain functional magnetic resonance image classification method based on a complex network, which comprises the following steps: pre-processing training sample images and test sample images, carrying out region segmentation, and extracting an average time sequence from each region; calculating the partial correlation coefficient among the average time sequences, carrying out matrix binarization on the partial correlation coefficient to obtain a complex network model, and calculating the feature path length, cost and clustering degree of the complex network model to respectively obtain network features of the training sample images and the test sample images; training to obtain an adaboost classifier; and by using the adaboost classifier obtained by training, classifying the test sample images. By using information in the brain functional magnetic resonance images as much as possible, the method can accurately classify the brain functional magnetic resonance images.

The invention discloses a method for adaptively detecting a remote obstacle in the technical field of robots. The method comprises the following steps of: acquiring an image and preprocessing the image; dividing area and performing super-pixel division processing; performing high-dimensional external feature extraction; obtaining topographic types of each super-pixel of a close scene area; obtaining low-dimensional leading features of the super-pixels of a close scene; obtaining the low-dimensional leading features of the super-pixels of a remote scene; performing Adaboost classifier training; and inputting the low-dimensional leading features of the super-pixels of the remote scene area into an Adaboost classifier, determining a current scene is the obstacle if the output of the Adaboost classifier is 1, and determining the current scene is the ground if the output of the Adaboost classifier is not 1. The method has the advantages of realizing the adaptive extraction of detection features of the obstacle, simplifying a classifier model, simultaneously reducing the influence of the multimode distribution of the obstacle and type ambiguity caused by random feature superposition on detection results, and improving obstacle detection accuracy and stability.

The invention discloses an Adaboost software defect unbalanced data classification method based on improvement, and mainly solves the problem that an existing software defect data classification method is poor in classification effect on minority classes. The method comprises the following steps that 1, software data is acquired from a software data set and then preprocessed, software module data is divided into a training set and a testing set for training and testing, and cross validation is performed for ten times; 2, feature selection of the software data is performed by combining a genetic algorithm based on improvement with a BP neural network to obtain an optimal feature subset, and then dimension reduction processing is performed on the software features; 3, the unbalancedness of the software defect data is fully considered, and an Adaboost classifier based on improvement is trained to classify software modules. According to the Adaboost software defect unbalanced data classification method based on improvement, the classification precision of the minority classes can be improved, and the software defect modules can be better detected.

The invention discloses a video-based flame detection method, and solves the technical problem in improving the flame detection accuracy. The video-based flame detection method comprises the followingsteps of obtaining a video image sequence; performing image preprocessing; performing color detection; performing flame feature extraction; and performing AdaBoost prediction. Compared with the priorart, the method has the advantages that all candidate flame points meeting a condition in a video image are found out according to characteristics of flame pixels in RGB and YCbCr color spaces; then,the video image is subjected to block segmentation; covariance matrixes corresponding to color and brightness attributes of flame pixel sets in video blocks are calculated; upper or lower triangularparts in the color and brightness covariance matrixes are extracted as eigenvectors; an obtained eigenvector set serves as an input of an AdaBoost classifier; the eigenvectors are input to an AdaBoostclassification model; an output is a judgment whether a fire happens or not; the fire happening situation can be detected out in real time; the false alarm rate can be reduced; and relatively high accuracy and strong robustness are achieved.

The invention discloses a traffic sign positioning method based on spherical panoramic video. Firstly, an Adaboost classifier of traffic signs is trained by Haar characteristics, and the traffic signs are detected on the basis; on the basis of detection results, centroid positions of detection regions are calculated, and the centroid positions are used as the positions of the traffic signs; two spherical live-action images containing the same traffic sign are chosen, and on the basis of the position mapping relation of the two spherical live-action images, the position of the traffic sign in three-dimensional space is determined; the coordinate of the spatial position is converted into a geodetic spatial rectangular coordinate, then the geodetic spatial rectangular coordinate is converted into a geodetic coordinate, longitude and latitude information in the geodetic coordinate is extracted, and positioning of the traffic signs in a two-dimensional map is achieved. By means of the traffic sign positioning method, users only need to drive on a road once, all surrounding scenes can be shot, therefore, positioning of all the traffic signs in the scenes can be performed, longitude and latitude coordinates corresponding to the signs on the live-action images can be calculated, and positioning is performed. Time and labor are saved, the positioning algorithm is high in efficiency, and real-time performance is good.

The invention discloses a multi-characteristic multi-model pedestrian detection method, comprising steps of using an ICF+Adaboost classifier A to process a video frame RGB image, using a foreground-mask-based pedestrian detection classifier to process the foreground mask, combining results of the two classifiers, dividing the results into a high confidence level pedestrian detection result and a low confidence level pedestrian detection result according to a threshold value, using an ICF+Adaboost classifier B and a DPM pedestrian detection classifier to perform respective detection on the low confidence level pedestrian detection result, combining detection results of the two classifiers, using a detection score, an overlapping ratio, a width-height ratio, a classifier sequence number and a foreground ratio of each detected pedestrian as characteristic vectors, inputting the characteristic vectors into a ruling SVM to determine whether the detected pedestrian is correct pedestrian detection, outputting a new pedestrian detection result and combining the new pedestrian detection result and the high confidence level pedestrian detection result into a set as a final detection result. The multi-characteristic multi-model pedestrian detection method effectively solves the problem in the prior art that the misjudgment rate is high, and improves the detection rate.

The invention discloses a face detection method and system based on image on-line learning. The face detection method based on image on-line learning comprises the steps of (1) preprocessing, wherein illumination compensation and graying processing are carried out on an image to be detected, image enhancement is carried out, nonlinear smooth filtering is carried out, denoising is carried out on the image, gray levels of pixels are normalized to obtain a high-quality gray level image, and then size normalization processing and edge detection processing are carried out; (2) carrying out face gesture detection, wherein the positions of the human eyes are determined, a human face area is divided, the rotation angle of a human face in the pitching dimension, the rotation angle of the human face in the depth dimension and the rotation angle of the human face in the plane dimension are detected, and whether the human face has an expression or not is automatically judged; (3) carrying out face detection, wherein the position of the human face in the image is determined, organs of the human face are located, and gray features of the image are selected, and are transmitted to a detection template which is trained in an off-line mode to carry out judgment; (4) updating, wherein the image which is processed through detection serves as a new sample to be applied to learning of a multi-layer cascade AdaBoost classifier, and the weight of set characteristic values of the multi-layer cascade AdaBoost classifier is updated. According to the face detection method and system based on image on-line learning, the accuracy rate of face detection can be improved.

The invention provides a gesture detection method based on multi-feature fusion. The method includes the steps that a cascaded Gentle Adaboost classifier is trained according to a method of fusion of the HOG feature, the variance feature and the Haar feature, so that a gesture classifier is formed; a skin color pre-examination is conducted through a skin color pre-examination module on an image collected by a camera and regions with the color similar to the skin color are screened out; the skin color regions which are screened out are traversed according to a sliding window method, and the input image considered to contain gestures is calibrated through rectangular frames; repeated rectangular region windows which are repeatedly judged as candidate gesture regions multiple times by the classifier are combined, and therefore the gesture of the image is calibrated. The method of fusion of the HOG feature, the variance feature and the Haar feature is selected for training the classifier and the image is measured through various feature values according to multiple peculiarities of the hands, so that the representing performance of the human hands is improved and the accuracy of a detection system is improved.

The invention provides a multi-face detecting and tracking method, belonging to the technical field of artificial intelligence. Firstly, according to multiple face characteristics in a monitoring system video, the method with the combination of Haar characteristic and an Adaboost classifier is used to detect faces, the concrete procedure comprises the following steps: (1) for a video continuous sequence, a Gaussian mixture model is used to extract moving foreground as a first class interest region, a region from the previous frame detection out of a motion region to a face as a center is taken as a second class interest region, and the face detection of the two interest regions is carried out, (2) a mean shift method is used to achieve multi-target tracking, and the adaptive updating and tracking of multiple faces of the same video are satisfied at the same time, (3) the multi-target tracking algorithm in the step (2) and the detection algorithm in the step (1) are combined, and a mixed multi-target tracking face detection algorithm is developed. According to the multi-face detecting and tracking method, the missed detection caused by face direction change or facial expression change is solved, the detection rate is raised, and the requirement of real-time performance by a monitoring system is satisfied.

The invention provides a human face detection method, comprising the steps of: importing: importing an image to be processed; sub-image extracting: traversing the image to be processed by using windows of different scales to extract parts of the image to be processed in the windows to serve as sub-images; preprocessing: specific to the sub-images, calculating a Sobel image of each sub-image by using a horizontal Sobel template, and excluding the sub-image if the ratio of the sum of Sobel response values of a predetermined region in the Sobel image to the total Sobel response value of the sub-image is smaller than a preset ratio threshold; detecting: specific to the sub-images passing the preprocessing, excluding non-human-face images by using an Adaboost cascade classifier to obtain candidate human-face sub-images; and verifying: specific to the candidate human-face sub-images, verifying the candidate human-face sub-images by using an Adaboost classifier established based on Haar-Sobel features so as to exclude the non-human-face sub-images, wherein the rest sub-images are taken as the human-face sub-images. The invention further provides human-face detection equipment correspondingly.

The invention discloses a method for detecting an adverse state of inclined sleeve part screws of a high-speed train overhead line system. The method comprises steps that firstly, a sample database of inclined sleeve parts is established, an AdaBoost classifier cascaded with HOG characteristic training of samples is extracted, and a supporting vector classifier is trained; secondly, Hough transformation is employed to realize extraction of an inclined sleeve inclination angle of a target image, and the inclination angle is made to rotate to a vertical direction; during fault determination, a bolt length and diameter ratio is taken as a criteria of a bolt loosing fault, and a relevant threshold is set to determine the bolt loosing fault; the bolt loosing fault is determined according to the position of a thin nut, differential processing on pixel accumulated distribution in a horizontal direction is carried out, and whether loosing occurs is determined according to a relevant horizontal pixel distribution change rate. Through the method, the state of inclined sleeve screw parts of the high-speed train overhead line system can be directly detected, an objective, true and accurate detection analysis result is acquired, and disadvantages of a traditional manual detection method are overcome.

The invention provides a specific color pedestrian detecting method in a static video camera scene. The method utilizes a Gaussian mixture model to conduct modeling to a background of the scene, after a background model is obtained, a sketchy motion foreground is obtained by comparing with a disposed image at present, a foreground is obtained by enabling the motion foreground to be raised in a rectangular block mode after filtering processing and morphological processing are conducted. A ratio occupied by foreground point in each foreground block is calculated and combines with statistical property of the projection number that the foreground points are projected on a horizontal coordinates to distinguish a complex foreground from a simple foreground, and a horizontal coordinates of a pedestrian which possibly exits is estimated. For the complex foreground block, a pedestrian height estimation model and an Adaboost classifier based on the feature of a direction gradient histogram are used for conducting accurate detection to the pedestrian, after a pedestrian target location is obtained, a clothing color of the pedestrian is identified. The specific color pedestrian detecting method in the static video camera scene improves detection speed and accuracy of the pedestrian target and can be applied to real-time video monitoring and video retrieval.

The invention discloses an embedded bottle pre-form detection system and a detection method. A bottle pre-form sidewall detection camera, a bottle opening thread side edge detection camera, a bottle opening size detection camera, a bottle opening thread top detection camera, a bottle opening sealing ring detection camera, and a bottle pre-form bottom detection camera are arranged on a detection line. Noise reduction, binarization and location fitting are carried out on the images through an image filtering algorithm based on morphology. The images are converted from an annular form to a strip form through a random sampling consistency algorithm. A registration operation is carried out on the images. Finally an Adaboost classifier is employed for defect identification. The embedded bottle pre-form detection system and the detection method have the advantages that the bottle blowing quality monitoring is transited from artificial detection to intelligent detection, the liquid state whole line loss is reduced by 5%, and the sealing fault ratio is reduced by 20%.