809results about "Character and pattern recognition" patented technology

The present invention relates to a vehicle telematics device with one or more cameras embedded in the same package for evidentiary surround audio-video recording, automatic accident detection and emergency help request notification, facial processing for drowsiness and distraction accident avoidance, and embedded multiple profiles of driver analytics, and mobile internet connectivity for contacting emergency services directly and without the delay of a call center and for mobile hot spot for in-vehicle infotainment. Each profile of driver analytics has many selectable advanced driver analysis parameters and multiple profiles can be active for a given driver. In case of emergency help request accident panoramic accident video is uploaded to a cloud drop box and link to video of accident is available along with severity index, location, nearest address, number of passengers, etc.

In some arrangements, product packaging is digitally watermarked over most of its extent to facilitate high-throughput item identification at retail checkouts. Imagery captured by conventional or plenoptic cameras can be processed (e.g., by GPUs) to derive several different perspective-transformed views—further minimizing the need to manually reposition items for identification. One claim recites a system for capturing imagery of retail items supported by a fixture, comprising: at least one camera unit having at least one operative camera to capture imagery of at least a portion of the retail items supported by the fixture; and one or more processors configured to process the captured imagery to discern identifying information associated with the imaged retail items. In one implementation the at least one camera unit comprises a plurality of cameras, wherein a field of view of a first one of the plurality of cameras overlaps with a field of view of a second one of the plurality of camera units. A great variety of other features, claims, combinations and arrangements are also detailed.

Described herein are systems and methods for performing multi-stage detection and classification of cancer regions from digitized images of biopsy slides. Novel methods for processing the digitized images to improve feature extraction and structure identification are disclosed, including but not limited to the use of quaternions, logarithmic mappings of color channels, and application of wavelets to logarithmic color channel mappings. The extracted features are utilized in improved machine learning algorithms that are further optimized to analyze multiple color channels in multiple dimensions. The improved machine learning algorithms include techniques for accelerating the training of the algorithms, making their application to biopsy detection and classification practical for the first time. The performance of the described systems and methods are further improved by the disclosure of a novel multistage machine learning scheme, in which additional classifiers are utilized to choose among the classes proposed by other classifiers in close cases.

Provided are a face recognition apparatus and method in which a facial image is divided into facial component images. The apparatus includes a component division unit that divides an input facial image into a plurality of facial component images; a face descriptor generating unit that generates face descriptors using a transform matrix corresponding to the respective facial component images, the face descriptors being characteristic vectors; a registered face descriptor database (DB) that stores registered face descriptors; and an authentication unit that authenticates the input facial image by comparing face descriptors for the input facial image input from the face descriptor generating unit with the registered face descriptors and providing predetermined weights corresponding to each facial component to the comparison results of each facial component.

A system for detecting intrusion into a space which includes motion detection light structure for producing and projecting a light pattern formed by a plurality of light beams of a certain character into a space and video motion detecting structure including a video camera having a field of view including at least some of the light pattern to detect intrusion into the light pattern. When motion is detected, images of the video camera are transmitted for threat assessment, and a scene illumination light is activated to improve visibility during nighttime operation.

The present invention relates to a method to make effective use of display space. In an embodiment of the invention, given a heterogeneous set of images along with metadata or nearby text, similar images are recursively clustered into a k-tree using the k-means algorithm. In an embodiment of the invention, the invention is particularly useful for showing image search results on small mobile devices.

A technique is disclosed that allows different computers in a network to create an identifier that uniquely identifies the network. The technique allows the unique identifier to be consistently created over time, regardless of the particular make up of the computing devices in the network at any particular point time. In some implementation, a computer within the network hosts the identification creation tool. In order to create a unique identifier for the network, the tool identifies each network adapter used by the host computer. Using this information, the tool identifies a gateway device used by the network adapter or adapters, and then determines the physical network address of that gateway device. For example, if the network is an Ethernet network, the tool will determine the medial access control (MAC) address for the gateway device. The tool then creates a unique identifier for the network based upon the physical address. The unique network identifier can then be used to associate the billing a usage fee, such as a licensing fee, with the network rather than with an individual computer within the network. Alternately or additionally, the unique network identifier can be used to maintain and support the gateway device for the network.

A gesture spotting detection method and apparatus employ a shoulder-line algorithm. The shoulder-line detecting method recognizes a GSD calling gesture that occurs in a shoulder-line, head or higher part in a remote distance or a short distance, although a user does not have a fixed posture. In the method, an image of people is received, and skin information of a person in the image is detected to detect a face area. Then, the cloth color information of the person is modeled from the inputted image to detect a cloth area. An external space is defined from the image based on the body space area, and an edge is extracted from the image based on the body space and the external space. Then, shoulder-line information is acquired based on an energy function obtained based on the body space, the external space, and the edge.

A correcting portion of an endoscope subjects a reference image to a magnification correction process so that the distance between a first mark and a second mark within the reference image is substantially equal to the distance between a first mark and a second mark within a comparative image. Next, the corrected reference image is subjected to a rotation correction process so that the orientation of the second mark with respect to the first mark within the reference image is substantially equal to the orientation of the second mark with respect to the first mark image within the comparison image. Further, the corrected reference image is subjected to a brightness distribution correction process so that the brightness distribution thereof becomes equal to that of the comparison image. The display state of the reference image and comparison image are made substantially equal, comparative diagnosis is facilitated, and diagnostic efficiency is improved.

The invention relates to an indoor scene semantic annotation method based on RGB-D data. According to the method, a coarse-to-fine global recursion feedback semantic annotation framework based on the RGB-D data is built, in addition, the whole semantic annotation framework is divided into two major parts including the coarse-granularity region stage semantic label deduction and fine-granularity pixel stage semantic label refinement. The framework is different from the traditional region stage or pixel stage semantic annotation framework, the framework rebuilds the relationship between the coarse-granularity region stage semantic label deduction and the fine-granularity pixel stage semantic annotation, and a reasonable global recursion feedback mechanism is introduced, so that the coarse-granularity region stage semantic annotation result and the fine-granularity pixel level semantic annotation result realize the alternate iterative updating optimization. Through adopting the mode, the multi-mode information of different region layers in the scene images is better merged, and the general problem that an annotation base element is difficult to be properly selected in the traditional indoor scene semantic annotation scheme is solved to a certain degree.

The present invention provides an industrial machine, which comprises an industrial machine main unit 14, a construction working implement 8 installed on the industrial machine main unit, two cameras 18 and 19 each mounted at a known position with respect to a machine center of the industrial machine main unit, at least three targets 21 being installed on the working implement so that the targets are within visual field of each of the two cameras, and an arithmetic processing device 23 for extracting the images of at least three targets in common to the images from stereo-images taken by the two cameras, for obtaining a three-dimensional position of each target image, and for calculating a position and a posture of the working implement with respect to the industrial machine main unit based on the three-dimensional position obtained.

The invention discloses a coal-rock identification method based on image gray level co-occurrence matrixes. The coal-rock identification method comprises the following steps of: respectively collecting color images f1 and f2 of a known coal sample and a known rock sample under the same imaging condition; respectively capturing sub-images s1 and s2 which are the same in size and free of a background; extracting gray level co-occurrence matrixes P1 and P2 of the sub-images; calculating image characteristics K1 and K2 based on the gray level co-occurrence matrixes; for an unknown coal-rock object to be identified, collecting an image fx under the same imaging condition; capturing a sub-image sx free of the background according to the same size; extracting a gray level co-occurrence matrix PXof the sx; calculating an image characteristic value Kx based on the gray level co-occurrence matrixes; and judging the coal-rock type according to a relation between Kx and K1 and K2. Because non-contact identification based on the images is utilized, the coal-rock identification method has the characteristics of easy deployment, strong adaptability, high identification rate and the like.

Techniques that facilitate tracking of physical locations of paper documents. Embodiments of the present invention can track physical locations of paper documents that have identification tags physically associated with them. Each identification tag that is physically associated with a paper document is configured to store information that can be used to uniquely identify the paper document with which the identification tag is physically associated. For a particular paper document, the identification tag physically associated with the particular paper document along with information stored in the identification tag facilitates tracking of the physical location of the particular paper document.

A method for recognizing objects in images is disclosed. The method of the present invention comprises the following steps. First, acquire a digital image. Then, select one or more objects from the image according to a certain characteristic. Next, generate an x-axis histogram and/or a y-axis histogram from the segmented image. Then, find the zeroes and maxima for the x-axis histogram and/or the y-axis histogram and use the polynomial regression analysis to determine the shape, shape and location of each of the objects in the segmented image according to the zeroes and maxima. If the two curves linking two zeroes and one maximum in the x-axis histogram and the y-axis histogram are two sloped line, the corresponding object may be determined to be a triangle. If each of the four curves linking two zeroes and two maxima is a line, the corresponding object may be determined to be a rectangle.

A user management method and apparatus are provided. The user management method includes verifying whether a user included in an input image is a registered user based on a first classification model classifying the registered user; storing a feature corresponding to the user and extracted from the input image in a database in response to the user not being the registered user; generating a second classification model classifying the registered user and a candidate user corresponding to one of features stored in the database; and determining whether the candidate user is to be registered based on the second classification model.

The invention provides a part size measurement method based on a charge coupled device (CCD). According to the part size measurement method, feature point recognition technology is adopted, pixels are classified through feature points of a part, the size of the whole part is measured, prior knowledge of the part is not required, and measurement for the common part is achieved. According to the method, high-speed measurement is conducted, and meanwhile a precision requirement is met. The part size measurement method based on the CCD is simple in algorithm, low in cost and good in application prospect due to the facts that only the CCD and a computer are required for hardware implementation and a complex mechanical structure is not required.

The invention discloses a segmentation method of pathological section unconventional cells based on a multi-scale hybrid segmentation model. The method comprises steps that positive and negative samples are respectively scaled to low-resolution, medium-resolution and high-resolution images, the full convolutional network algorithm is utilized for training to acquire a convergent low-resolution segmentation model, a medium-resolution segmentation model and a high-resolution model; a multi-scale hybrid segmentation model is acquired through fusion by a model integration method; after the effective discriminating area of a new pathological section is processed through utilizing the data enhancement method during testing, the processed effective discriminating area is inputted to the multi-scale hybrid segmentation model, the probability of each pixel in the effective segmentation area is outputted, pixels with probability values greater than the threshold t are taken as abnormal cell pixels and are recorded as 1, remaining pixels are taken as normal cell pixels and are recorded as 0, binary images predicted by the multi-scale hybrid segmentation model are acquired, and post-processingon the binary images is carried out to acquire the final segmentation result. The method is advantaged in that high precision is realized, and a Dice value is above 0.869.

The invention discloses a behavior recognition method based on intelligent sub-space networks. The method specifically includes the following steps of preprocessing and learning spatial characteristics, extracting spatial and temporal characteristics, representing videos as histogram vectors, training SVM classifiers and recognizing behaviors. According to the behavior recognition method based on the intelligent sub-space networks, the spatial characteristics are extracted from video data, the spatial characteristics are then pooled so that the characteristics equivalent to the spatial and temporal characteristics can be obtained, each video sequence is represented as the corresponding histogram vector about all words in a word bag through a word bag method, multiple types of behaviors are recognized through the trained SVM classifiers and a one-to-many strategy, the robustness of behavior recognition is increased, and the behavior recognition rate is increased.

The invention discloses a characteristics matching and tracking based video monitoring apparatus and a method. In the invention, high definition cameras must be installed to cover a certain area. The mutually connected high definition cameras are then uniformly connected to a switch, with the consequence of forming a monitoring network so that videos from the cameras can be transmitted from one to another. When a person enters such an area, an HOG algorithm is adopted to monitor the video streams in real time; the targets monitored under the HOG algorithm are listed in numbers with each moving target provided with an identifier of its own and its location known all the time. The characteristics of the targets in the videos are then extracted and analyzed to decide if these targets behave in normal or not. If their behaviors are normal ones, then these targets are not trouble makers. Otherwise, an alarm control device sends a warning to draw the attention of a security man. At the time, the security man can track and observe the behaviors of the persons from the monitoring images and promptly deal with this, therefore preventing something even worse happening.

The invention discloses an obstacle feature extraction method based on a millimeter wave radar and a laser radar. The method includes the following steps that: step 1, radar raw data are acquired; step 2, the radar original data are solved according to a radar protocol, so that obstacle position information can be obtained; step 3, the solved millimeter wave radar data are screened; step 4, time and space synchronization is performed on the millimeter wave radar and the laser radar; step 5, regions of interest (ROI) are established in a laser radar coordinate system according to millimeter wave radar information and are denoted as Ri, wherein i is equal to 1, 2, ... n, wherein n is the number of the effective targets of the millimeter wave radar; step 6, the point cloud data of the i-th ROI are processed according to a scattered point cloud data simplification processing method; and step 7, a two-dimensional image is generated on the basis of the point cloud data, edge detection is performed on the image, and obstacle features are extracted. With the method adopted, obstacle edge information can be obtained, and therefore, more references can be provided for a driverless decision-making layer.

An apparatus and method for providing an augmented reality (AR) user interface. The method includes acquiring an image containing at least one object; recognizing the object from the acquired image; detecting AR information related to the recognized object; classifying the detected AR information into groups according to specific property information; generating a user interface that displays the groups of AR information separately.

The present invention relates to a pedestrian detection method based on deep learning and multi-feature point fusion. The pedestrian detection method is characterized by at a training stage, firstly acquiring a pedestrian image under an application scene, marking the head and shoulder parts of the pedestrians in the image, and then using the pedestrian samples for the model training, wherein the model training comprises two steps of 1) taking the head and shoulder images of the pedestrians as the training samples, training a dichotomy model of the head and shoulder parts of the pedestrians; 2) using the model parameters obtained by the training in the step 1) to initialize partial parameters of a pedestrian detection model in a transfer learning manner. The pedestrian detection method of the present invention can overcome the problem that the pedestrians shield mutually to a certain extent, adopts a deep learning method to extract the pedestrian features, can better overcome the actual application problem that the factors, such as the pedestrian clothing, postures, backgrounds, illumination conditions, etc., change, also can effectively overcome the problems of the pedestrian multiple postures, the pedestrian multiple scales, the pedestrian mutual shielding, etc., and enables the pedestrian detection accuracy and robustness to be improved substantially.

The invention is suitable for the technical field of communication, and provides an identity recognition method. The method comprises the following steps of: storing the original fingerprint information of a user; randomly sending a random fingerprint verification instruction after receiving an identity recognition request of the user; acquiring fingerprint information input by the user according to the random fingerprint verification instruction; and judging whether the fingerprint information input by the user is matched with the random fingerprint verification instruction and the original fingerprint information simultaneously, if so, indicating that identity recognition is successful, and otherwise, indicating that identity recognition fails. Correspondingly, the invention also provides an identity recognition system and an identity recognition terminal. Therefore, a safer and more reliable identity recognition technology can be provided, and the implementation mode is simpler.