WO2019104953A1 - Procédé et appareil de localisation et terminal mobile - Google Patents

Procédé et appareil de localisation et terminal mobile Download PDF

Info

Publication number
WO2019104953A1
WO2019104953A1 PCT/CN2018/087337 CN2018087337W WO2019104953A1 WO 2019104953 A1 WO2019104953 A1 WO 2019104953A1 CN 2018087337 W CN2018087337 W CN 2018087337W WO 2019104953 A1 WO2019104953 A1 WO 2019104953A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
positioning
feature
scene
image feature
Prior art date
Application number
PCT/CN2018/087337
Other languages
English (en)
Chinese (zh)
Inventor
简伟华
Original Assignee
深圳市智能现实科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市智能现实科技有限公司 filed Critical 深圳市智能现实科技有限公司
Publication of WO2019104953A1 publication Critical patent/WO2019104953A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames

Definitions

  • the present invention relates to the field of computer application technologies, and in particular, to a positioning method and device, and a mobile terminal.
  • Bluetooth iBeacon, Wifi, etc. are mainly used for positioning.
  • the user connects to the nearest iBeacon device via Bluetooth to estimate the real-time location and display it on the floor plan.
  • the accuracy is not more than a few meters and is not intuitive enough, and the coverage of a single iBeacon The range is limited, resulting in blind spots and offsets; when locating via Wifi, Wifi devices need to be deployed in advance, and AP (Wireless Access Point) relay is used.
  • Users can locate their real-time location by connecting to WiFi.
  • the accuracy error is large (possibly at ten meters), and the user needs to log in to connect to a specific WiFi when positioning, which results in a cumbersome positioning operation.
  • the present invention provides a positioning method and apparatus, and a mobile terminal.
  • a positioning method comprising:
  • a positioning device comprising:
  • a positioning photo acquisition module configured to acquire a positioning photo of the location of the scene in the preset scene in the mobile terminal
  • a metric learning module configured to perform metric learning on the image features in the positioning photo and the feature index, and calculate a similarity between the positioning photo and each image feature, where the feature index is collected from the preset scene An index of association between image features extracted in the panoramic video and corresponding scene positions in the preset scene;
  • a target image feature determining module configured to determine, according to the similarity, a target image feature that is most similar to the positioning photo
  • a positioning module configured to determine a scene position associated with the target image feature from a scene position of the feature index, and locate a panoramic image frame corresponding to the target scene position in the panoramic video.
  • a mobile terminal including:
  • the memory stores readability instructions that, when executed by the processor, implement the method of the first aspect.
  • a computer readable storage medium having stored thereon a computer program that, when executed, implements the method of the first aspect.
  • the positioning photos collected by the scene position of the preset scene in the mobile terminal are acquired, and the image features in the positioning photo and the feature index are measured and learned, and the positioning photo and each image feature are calculated. Similarity, and determining the target image feature most similar to the positioning photo according to the similarity degree, determining the scene position associated with the target image feature from the scene position of the feature index, and locating the panoramic image corresponding to the target scene position in the panoramic video
  • the frame can be used to locate the corresponding panoramic image frame of the scene position in the panoramic video by the positioning photo acquired at the scene position, thereby realizing high-precision positioning of the scene position easily.
  • FIG. 1 is a flow chart of a positioning method according to an exemplary embodiment.
  • FIG. 2 is a flow chart showing another positioning method according to the corresponding embodiment of FIG. 1.
  • FIG. 3 is a specific implementation flowchart of step S210 in the positioning method according to the corresponding embodiment of FIG. 2 .
  • FIG. 4 is a specific implementation flowchart of step S130 in the positioning method according to the corresponding embodiment of FIG. 1.
  • FIG. 5 is a flow chart showing another specific implementation of step S130 in the positioning method according to the corresponding embodiment of FIG. 1.
  • FIG. 6 is a block diagram of a positioning device, according to an exemplary embodiment.
  • Figure 7 is a block diagram of another positioning device illustrated in accordance with the corresponding embodiment of Figure 6.
  • FIG. 8 is a block diagram of an image feature extraction module 210, illustrated in accordance with the corresponding embodiment of FIG.
  • FIG. 9 is a block diagram of a target image feature determination module 130 shown in accordance with the corresponding embodiment of FIG.
  • FIG. 10 is another block diagram of the target image feature determination module 130 shown in accordance with the corresponding embodiment of FIG.
  • FIG. 11 is a block diagram of a mobile terminal according to an exemplary embodiment.
  • FIG. 1 is a flowchart of a positioning method according to an exemplary embodiment.
  • the method may be run on a mobile terminal, or may be run on a server that performs data transmission with a mobile terminal.
  • the positioning method may be Includes the following steps.
  • step S110 a positioning photo collected in the mobile terminal for the scene position in the preset scene is acquired.
  • a positioning photo is a photo taken at a certain scene position of a preset scene. For example, when the user presets the scene position A of the scene, the user collects the photos through the camera of the mobile terminal.
  • the positioning photo may be collected on the spot of the scene in the preset scene when the user is positioned, or may be stored in the mobile terminal in advance in the preset scene, or may be stored in the preset scene in other manners.
  • the location of the scene is used to collect the captured photos.
  • the mobile terminal may be a computer device such as a mobile phone or a tablet computer, and the specific implementation manner is not limited by this embodiment.
  • the preset scene is an indoor scene, but the preset scene may also be an outdoor scene, and the specific form of the preset scene is not limited herein.
  • step S120 the locating photos and the image features in the feature index are subjected to metric learning, and the similarity between the locating photos and the respective image features is calculated.
  • the feature index is an association index between image features extracted from the panoramic video collected for the preset scene and corresponding scene positions in the preset scene.
  • Each image feature corresponds to a corresponding scene position in the preset scene.
  • the panoramic video is a comprehensive video captured in advance for a preset scene.
  • a panoramic video is composed of multiple consecutive panoramic image frames.
  • the panoramic video is different from the normal video.
  • the image can be displayed in any viewing direction of a certain scene position in the preset scene associated with the panoramic video.
  • the image feature is high-dimensional data extracted from the panoramic video and associated with the image information.
  • the image features are first extracted from the positioning photos, and then the image features extracted from the positioning photos and the image features in the feature index are metrically learned.
  • Metric learning is the calculation of the similarity of image features in the feature index after extracting image features from the positioned photos.
  • distance function For example, if our goal is to recognize faces, then we need to build a distance function to enhance the appropriate features (such as hair color, face, etc.); and if our goal is to recognize gestures, then we need to build a capture pose similarity. Distance function. To handle a wide variety of similarities, we can build distance functions by selecting the appropriate features for a particular task.
  • step S130 the target image feature most similar to the positioned photo is determined based on the similarity.
  • the image feature with the highest similarity between the captured photos may be selected as the target image feature; and the similarity between each image feature and the positioned photo may also be adopted. Comparing with the preset similarity threshold, respectively, and then re-selecting the target image feature from the image feature with the similarity between the positioning photos and the similarity threshold; and determining the similarity to the positioning photo according to the similarity by other means.
  • Target image features are provided.
  • step S140 the scene position associated with the target image feature is determined from the scene position of the feature index, and the panoramic image frame corresponding to the scene position is located in the panoramic video.
  • the panoramic video is an omnidirectional video collected for a preset scene
  • the panoramic image frame of the panoramic video corresponds to the scene position in the preset scene
  • the image feature is the preset scene.
  • the location of the scene is associated.
  • the scene position associated with the target image feature is determined from the feature index by the target image feature, and then the panoramic image corresponding to the scene position is located in the panoramic video according to the scene position.
  • the frame is finally positioned and navigated by the user through the playback of the panoramic video.
  • the panoramic image frame of the panoramic video is used to perform high-precision positioning of the scene position of the positioning photo collection.
  • the positioning photos collected in the preset scene in the mobile terminal are acquired, and the image features in the positioning photo and the feature index are measured and learned, and the positioning photos and each are calculated.
  • the similarity between the image features, and determining the target image feature most similar to the positioned photo according to the similarity, and determining the scene position associated with the target image feature from the scene position of the feature index, and positioning the target scene in the panoramic video The panoramic image frame corresponding to the position, so that the user can locate the corresponding panoramic image frame in the panoramic video by simply positioning the captured photo by the scene position in the preset scene, so as to easily and accurately locate the scene position. .
  • FIG. 2 is a flow chart showing another positioning method according to the corresponding embodiment of FIG. 1. As shown in FIG. 2, after step S120 in the positioning method shown in FIG. 1, the positioning method may further include the following steps.
  • step S210 image features of each panoramic image frame are extracted by performing deep learning on each panoramic image frame of the panoramic video.
  • Deep learning is a method based on the representation of data in machine learning.
  • CNN Convolutional Neural Network
  • RNN Recurrent Neural Network
  • NBN DBN
  • VGG13 network VGGNET network (other network) does not limit the specific method of deep learning here.
  • deep learning is performed through CNN, which is a deep learning network and an artificial network.
  • the network uses a weight sharing method. More similar to biological neural networks. This design not only reduces the complexity of the model, but also greatly reduces the number of network weights.
  • convolutional neural networks high-dimensional data files can be processed directly (because the image features we extract are high-dimensional data) without the need to extract features and data from panoramic image frames as in traditional neural network algorithms. Rebuilding, and often these processes require very large overhead. Therefore, in convolutional neural networks, the training speed of the model is very fast compared to the traditional neural network. Through the construction of the multi-layer perceptron in the convolutional neural network, some high-dimensional data with different structural translations and high-dimensional data with different dimensions can be highly tolerated.
  • CNN uses the method of weight sharing to reduce the number of parameters that need to be learned. Compared with the general forward BP algorithm (Error Back Propagation), the training speed and accuracy are greatly improved. As a deep learning algorithm, CNN can minimize the overhead of data preprocessing. In the algorithm, the input of the lowest layer as the hierarchical structure is only a small part of the high-dimensional data (that is, the local sensing area or the visual field sensor), and the processed information is transmitted to a deeper level, and each layer is filtered. To obtain the characteristics of the observation data input by the upper layer.
  • the panoramic video is an omnidirectional video captured on a preset scene
  • the panoramic image frame of the panoramic video is an omnidirectional image acquired at a certain scene position.
  • the similarity calculation is performed directly on the positioning photo and the panoramic image frame, the calculation amount is greatly increased, and the processing time of the similarity calculation is increased. Therefore, through deep learning, the image features of each panoramic image frame are extracted, and the difference between different panoramic image frames is enlarged, thereby reducing the calculation amount when the similarity calculation is performed between the positioning photo and the panoramic image frame, and the similarity calculation is shortened. Processing time.
  • step S220 for each image feature of the panoramic image frame, a feature index is established between each image feature of the panoramic image frame and the corresponding scene position according to the scene position corresponding to the panoramic image frame.
  • the panoramic image frame is in one-to-one correspondence with the scene position in the preset scene.
  • an association index between the image features of each panoramic image frame and the scene position is established.
  • the scene position associated with the target image feature can be quickly located.
  • an image feature is extracted from the panoramic video collected for the preset scene by depth learning in advance, and an association index between the image feature and the corresponding scene position in the preset scene is established.
  • the position of the scene associated with the target image feature can be quickly located, which greatly reduces the amount of data calculation during positioning, shortens the time-consuming positioning, and improves the positioning efficiency.
  • FIG. 3 is a detailed description of step S210 in the positioning method illustrated in FIG. 2 corresponding to the exemplary embodiment.
  • the playback control command may include a playback direction control command
  • the step S210 may include the following steps.
  • step S211 for each panoramic image frame of the panoramic video, the panoramic image frame is subjected to image segmentation according to a preset angle to obtain a segmented image sequence.
  • the panoramic image frame of the panoramic video is an omnidirectional image.
  • the positioning photo acquired at the scene position in the preset scene is a plane photo.
  • each panoramic image frame is divided into divided images that are close to the size of the positioned image, thereby further improving the accuracy of the similarity calculation.
  • Image segmentation is the cutting of a panoramic video image frame, and each panoramic video image frame is cut into a plurality of planar images.
  • the horizontal direction is 45 degrees
  • the pitch angle is 60 degrees
  • the horizontal is divided into 10 divided images every 36 degrees
  • 10 divided images are vertically divided every 24 degrees
  • a total of 40 divided images are intercepted.
  • the divided image sequence is a divided image set obtained by performing image segmentation on a panoramic image frame, and the divided image sequence includes a plurality of divided images.
  • step S212 depth learning is performed on each of the divided images in the divided image sequence by the depth learning algorithm, and image features of the respective divided images are extracted.
  • the panoramic image frame in the panoramic video is first image segmented, and then each segment image is extracted.
  • the image feature further calculates the similarity between the image features extracted from the positioned photos and the image features of the divided images. Since the positioned photos and the segmented images are both planar images and the image sizes are close, the accuracy of the similarity calculation is greatly improved. Degree, the data processing amount of the similarity calculation is reduced, the positioning time is shortened, and the positioning efficiency is improved.
  • step S130 in FIG. 1 may include the following steps.
  • step S131 by comparing with a preset similarity threshold, the to-be-selected image feature whose similarity with the positioned photo reaches the similarity threshold is determined.
  • the image feature to be selected is an image feature having a higher degree of similarity with the positioned photo.
  • the preset similarity threshold is a similarity threshold set in advance. If the similarity between the image feature and the positioned photo does not reach the similarity threshold, it indicates that the image feature does not match the positioned photo.
  • the target image features most similar to the positioned photos are determined in the bloody image features.
  • step S132 the target image feature most similar to the positioned photo is determined from the candidate image features by the classification network.
  • a classification network is a network that determines whether two image features are similar.
  • the classification network adopts the Matchnet framework
  • the network structure may be Alexnet, VGG, Googlenet, Resnet, and the like.
  • Alexnet is a CNN-type deep learning network model.
  • Matchnet is a network of two branches. Each branch network structure uses Alexnet. The input is a pair of images. The output is whether the two images are similar. .
  • the network structure therein is not limited by the exemplary embodiment.
  • the method as described above by using a preset similarity threshold, when there are a plurality of image features whose similarity with the positioned photo reaches the similarity threshold, determining and positioning the photo from the plurality of image features through the classification network
  • the most similar target image features avoid erroneously determining the image features most similar to the positioned photos, thereby greatly improving the accuracy of determining the target image features most similar to the positioned photos, and greatly improving the accuracy of positioning by positioning the photos.
  • FIG. 5 is a detailed description of step S130 in the positioning method illustrated in FIG. 1 corresponding to the exemplary embodiment. As shown in FIG. 5, the number of the positioning photos is multiple, and the plurality of positioning photos are photos taken in different viewing directions of the scene position, and step S130 in FIG. 1 may include the following steps.
  • step S135 for each image feature, the candidate probability of the image feature is calculated based on the similarity between the image feature and each of the positioned photos acquired at the scene location.
  • the panoramic video is omnidirectional, and the scene position of the preset scene also includes multiple viewing directions.
  • the viewing direction is the display direction of the panoramic image frame in the panoramic video when the associated preset scene is displayed, and corresponds to the visual field orientation when the corresponding scene position in the preset scene associated with the panoramic video.
  • the candidate probability is a probability size that determines that the image feature is the most similar target image feature to the positioned photo.
  • the image features and the positioning photos in the feature index are respectively measured and analyzed, and the image features and the positioning photos in the feature index are respectively calculated.
  • the degree of similarity between the image features and the scene position in the feature index is calculated according to the similarity, so that the similarity between the image features in the plurality of positioning photos and the feature index is mutually verified, thereby further improving the determination.
  • the accuracy of the target image features is calculated according to the similarity, so that the similarity between the image features in the plurality of positioning photos and the feature index is mutually verified, thereby further improving the determination.
  • the candidate probability of the image feature can be calculated according to the similarity between the image feature and each of the positioned photos of the scene position in various ways.
  • the average degree of similarity may be calculated according to the similarity between the image feature and each of the positioning photos collected at the scene position to obtain the candidate probability of the image feature; or the preset similarity may be adopted. Threshold, only when the similarity between the image feature and each of the positioned photos collected at the scene position reaches the similarity threshold, the average value is calculated according to the similarity between the image feature and each of the positioned photos, and the image feature is obtained.
  • the probability of the candidate to be selected; the similarity between the image feature and each of the locating photos may be sorted according to the degree of similarity, and then the weights of the image features may be calculated by setting different weights for each sorting; or may be based on other methods.
  • the degree of similarity between the image features and the respective positioned photos of the scene position, and the candidate probability of the image features are calculated, and are not described one by one.
  • step S136 based on the candidate probability, the target image feature most similar to the location photo acquired at the scene location is determined.
  • the image features corresponding to the maximum candidate probability are determined to be the most similar target image features.
  • the locating photos collected in different viewing directions of the scene position are respectively metrically analyzed with the image features in the feature index, thereby performing the similarity between the image features in the plurality of locating photos and the feature index.
  • Mutual verification further improve the accuracy of determining the characteristics of the target image.
  • FIG. 6 is a block diagram of a positioning apparatus, including but not limited to: a positioning photo acquisition module 110, a metric learning module 120, a target image feature determination module 130, and a positioning module 140, according to an exemplary embodiment.
  • the positioning photo acquisition module 110 is configured to acquire a positioning photo of the location of the scene in the preset scene in the mobile terminal;
  • the metric learning module 120 is configured to perform metric learning on the image features in the positioning photo and the feature index, and calculate a similarity between the positioning photo and each image feature, where the feature index is from the preset scene An association index between the image features extracted in the collected panoramic video and corresponding scene positions in the preset scene;
  • the target image feature determining module 130 is configured to determine, according to the similarity, a target image feature that is most similar to the positioning photo;
  • the positioning module 140 is configured to determine a scene position associated with the target image feature from a scene position of the feature index, and locate a panoramic image frame corresponding to the target scene position in the panoramic video.
  • the positioning apparatus shown in FIG. 6 further includes, but is not limited to, an image feature extraction module 210 and a feature index establishment module 220.
  • the image feature extraction module 210 is configured to extract image features of each panoramic image frame by performing deep learning on each panoramic image frame of the panoramic video.
  • the feature index establishing module 220 is configured to establish a feature between each image feature of the panoramic image frame and the corresponding scene position according to the image position corresponding to the panoramic image frame for the image feature of each panoramic image frame. index.
  • the image feature extraction module 210 illustrated in FIG. 7 includes, but is not limited to, an image segmentation unit 211 and an image feature extraction unit 212 .
  • the image segmentation unit 211 is configured to perform image segmentation on the panoramic image frame according to a preset angle for each panoramic image frame of the panoramic video to obtain a segmentation image sequence.
  • the image feature extraction unit 212 is configured to perform depth learning on each of the divided image sequences by using a depth learning algorithm, and extract image features of each divided image.
  • the target image feature determining module 130 illustrated in FIG. 6 includes, but is not limited to, a desired control direction determining unit 131 and a target image feature determining unit 132 .
  • the desired control direction determining unit 131 is configured to use a candidate image feature determining unit, configured to determine, by comparison with a preset similarity threshold, a candidate image that has a similarity with the positioning photo to reach the similarity threshold. feature;
  • the first target image feature determining unit 132 is configured to determine, from the candidate image features, a target image feature that is most similar to the positioning photo by using a classification network.
  • the number of the positioning photos is multiple, and the plurality of positioning photos are photos acquired in different viewing directions of the scene position, and the target image feature determining module 130 illustrated in FIG. 6 includes It is not limited to: the candidate probability calculation unit 135 and the target image feature determination unit 136.
  • the candidate probability calculation unit 135 is configured to calculate, for each image feature, a candidate probability of the image feature according to a similarity between the image feature and each of the positioning photos collected at the scene location;
  • the second target image feature determining unit 136 is configured to determine, according to the candidate probability, a target image feature that is most similar to the positioning photo collected at the scene location.
  • FIG. 11 is a block diagram of a mobile terminal 100, according to an exemplary embodiment.
  • the mobile terminal 100 may include one or more of the following components: a processing component 101, a memory 102, a power component 103, a multimedia component 104, an audio component 105, a sensor component 107, and a communication component 108.
  • the above components are not all necessary, and the mobile terminal 100 may add other components or reduce some components according to the requirements of the functions, which is not limited in this embodiment.
  • Processing component 101 typically controls the overall operations of mobile terminal 100, such as operations associated with displays, telephone calls, data communications, camera operations, and recording operations, and the like.
  • Processing component 101 can include one or more processors 109 to execute instructions to perform all or part of the steps described above.
  • processing component 101 can include one or more modules to facilitate interaction between component 101 and other components.
  • processing component 101 can include a multimedia module to facilitate interaction between multimedia component 104 and processing component 101.
  • the memory 102 is configured to store various types of data to support operations at the mobile terminal 100. Examples of such data include instructions for any application or method operating on the mobile terminal 100.
  • the memory 102 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as SRAM (Static Random Access Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory) Erase programmable read-only memory, EPROM (Erasable Programmable Read Only Memory), PROM (Programmable Read-Only Memory), ROM (Read-Only Memory) Read memory), magnetic memory, flash memory, disk or optical disk. Also stored in memory 102 is one or more modules configured to be executed by the one or more processors 109 to perform any of Figures 1, 2, 3, 4, and 5. All or part of the steps in the method shown.
  • Power component 103 provides power to various components of mobile terminal 100.
  • Power component 103 can include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power for mobile terminal 100.
  • the multimedia component 104 includes a screen that provides an output interface between the mobile terminal 100 and a user.
  • the screen may include an LCD (Liquid Crystal Display) and a TP (Touch Panel). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from the user.
  • the touch panel includes one or more touch sensors to sense touches, slides, and gestures on the touch panel. The touch sensor may sense not only the boundary of the touch or sliding action, but also the duration and pressure associated with the touch or slide operation.
  • the audio component 105 is configured to output and/or input an audio signal.
  • the audio component 105 includes a microphone that is configured to receive an external audio signal when the mobile terminal 100 is in an operational mode, such as a call mode, a recording mode, and a voice recognition mode.
  • the received audio signal may be further stored in memory 102 or transmitted via communication component 108.
  • the audio component 105 also includes a speaker for outputting an audio signal.
  • Sensor component 107 includes one or more sensors for providing various aspects of state assessment for mobile terminal 100.
  • the sensor component 107 can detect an open/close state of the mobile terminal 100, the relative positioning of the components, and the sensor component 107 can also detect a change in coordinates of a component of the mobile terminal 100 or the mobile terminal 100 and a temperature change of the mobile terminal 100.
  • the sensor assembly 107 can also include a magnetic sensor, a pressure sensor, or a temperature sensor.
  • Communication component 108 is configured to facilitate wired or wireless communication between mobile terminal 100 and other devices.
  • the mobile terminal 100 can access a wireless network based on a communication standard such as WiFi (WIreless-Fidelity), 2G or 3G, or a combination thereof.
  • communication component 108 receives broadcast signals or broadcast associated information from an external broadcast management system via a broadcast channel.
  • the communication component 108 further includes an NFC (Near Field Communication) module to facilitate short range communication.
  • the NFC module can be based on RFID (Radio Frequency Identification) technology, IrDA (Infrared Data Association) technology, UWB (Ultra-Wideband) technology, BT (Bluetooth) technology and others. Technology to achieve.
  • the mobile terminal 100 may be configured by one or more ASICs (Application Specific Integrated Circuits), DSP (Digital Signal Processing), and PLD (Programmable Logic Device). Device), FPGA (Field-Programmable Gate Array), controller, microcontroller, microprocessor or other electronic component implementation for performing the above method.
  • ASICs Application Specific Integrated Circuits
  • DSP Digital Signal Processing
  • PLD Programmable Logic Device
  • FPGA Field-Programmable Gate Array
  • controller microcontroller
  • microprocessor microprocessor or other electronic component implementation for performing the above method.
  • the present invention further provides a mobile terminal, which performs all or part of the steps of the positioning method shown in any of FIG. 1, FIG. 2, FIG. 3, FIG. 4 and FIG.
  • the device includes:
  • the memory stores readability instructions that, when executed by the processor, implement a method as described in any of the above-described exemplary embodiments.
  • a storage medium is also provided, which is a computer readable storage medium, such as a temporary and non-transitory computer readable storage medium including instructions.
  • the storage medium includes, for example, a memory 102 of instructions that are executable by the processor 109 of the mobile terminal 100 to perform the positioning method described above.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)

Abstract

La présente invention se rapporte au domaine technique des applications informatiques. L'invention porte sur un procédé et sur un appareil de localisation, ainsi que sur un terminal mobile. Ledit procédé comprend les étapes consistant : à acquérir une photographie à localiser, capturée par un terminal mobile, d'un emplacement dans une scène prédéfinie ; à effectuer un apprentissage de mesure sur la photographie à localiser et sur des caractéristiques d'image contenues dans un index de caractéristiques, et à calculer une similarité entre la photographie à localiser et chaque caractéristique d'image, l'index de caractéristiques étant un index concernant l'association entre les caractéristiques d'image extraites d'une vidéo panoramique capturée pour la scène prédéfinie et des emplacements correspondants dans la scène prédéfinie ; en fonction de la similarité, à déterminer la caractéristique d'image cible la plus similaire à la photographie à localiser ; et à déterminer, parmi les emplacements dans l'index de caractéristiques, un emplacement associé à la caractéristique d'image cible, et à localiser, dans la vidéo panoramique, une trame d'image panoramique correspondant à l'emplacement. Ledit procédé et ledit appareil de localisation, ainsi que le terminal mobile, peuvent facilement et commodément localiser un emplacement avec une précision élevée.
PCT/CN2018/087337 2017-12-01 2018-05-17 Procédé et appareil de localisation et terminal mobile WO2019104953A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711258377.7 2017-12-01
CN201711258377.7A CN108009588A (zh) 2017-12-01 2017-12-01 定位方法及装置、移动终端

Publications (1)

Publication Number Publication Date
WO2019104953A1 true WO2019104953A1 (fr) 2019-06-06

Family

ID=62055952

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/087337 WO2019104953A1 (fr) 2017-12-01 2018-05-17 Procédé et appareil de localisation et terminal mobile

Country Status (2)

Country Link
CN (1) CN108009588A (fr)
WO (1) WO2019104953A1 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110852153A (zh) * 2019-09-26 2020-02-28 深圳市金溢科技股份有限公司 一种泊位状态检测方法、获取方法、装置、工控机及介质
CN111314739A (zh) * 2020-02-17 2020-06-19 聚好看科技股份有限公司 一种图像处理方法、服务器及显示设备
CN111582378A (zh) * 2020-05-09 2020-08-25 上海钧正网络科技有限公司 定位识别模型的训练生成方法、位置检测方法、装置

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108009588A (zh) * 2017-12-01 2018-05-08 深圳市智能现实科技有限公司 定位方法及装置、移动终端
WO2020019239A1 (fr) * 2018-07-26 2020-01-30 深圳前海达闼云端智能科技有限公司 Procédé et dispositif de positionnement, terminal et support de stockage lisible
CN109359517A (zh) * 2018-08-31 2019-02-19 深圳市商汤科技有限公司 图像识别方法和装置、电子设备、存储介质、程序产品
CN109671119A (zh) * 2018-11-07 2019-04-23 中国科学院光电研究院 一种基于slam的室内定位方法及装置
CN111291588A (zh) * 2018-12-06 2020-06-16 新加坡国立大学 用于在建筑物内定位的方法和系统
CN112074062A (zh) * 2019-05-21 2020-12-11 广东小天才科技有限公司 一种基于场景的灯光调节方法及智能照明装置
CN110276841B (zh) * 2019-06-27 2023-11-24 北京小米移动软件有限公司 应用于增强现实设备的运动轨迹确定方法、装置及终端
WO2021237443A1 (fr) * 2020-05-26 2021-12-02 蜂图志科技控股有限公司 Procédé et appareil de positionnement visuel, dispositif et support d'enregistrement lisible
CN112860946B (zh) * 2021-01-18 2023-04-07 四川弘和通讯集团有限公司 一种视频图像信息转换为地理信息的方法及系统
CN114353796A (zh) * 2021-12-06 2022-04-15 煤炭科学技术研究院有限公司 一种室内定位方法、装置及电子设备

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20010003182A1 (en) * 1999-12-03 2001-06-07 Lilian Labelle Method and devices for indexing and seeking digital images taking into account the definition of regions of interest
CN106909684A (zh) * 2017-03-06 2017-06-30 青岛鹏海软件有限公司 全景图像检索及展示方法
CN108009588A (zh) * 2017-12-01 2018-05-08 深圳市智能现实科技有限公司 定位方法及装置、移动终端

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102918858B (zh) * 2010-12-24 2014-09-03 富士胶片株式会社 立体全景图像创建设备和方法、立体全景图像再现设备和方法
CN103162682B (zh) * 2011-12-08 2015-10-21 中国科学院合肥物质科学研究院 基于混合现实的室内路径导航方法
CN103106252A (zh) * 2013-01-16 2013-05-15 浙江大学 利用手持设备的平面区域定位方法
CN103198488B (zh) * 2013-04-16 2016-08-24 北京天睿空间科技有限公司 Ptz监控摄像机实时姿态快速估算方法
CN104748738B (zh) * 2013-12-31 2018-06-15 深圳先进技术研究院 室内定位导航方法和系统
CN106095881A (zh) * 2016-06-07 2016-11-09 惠州Tcl移动通信有限公司 一种显示照片相应信息的方法、系统及移动终端
CN106358033B (zh) * 2016-08-25 2018-06-19 北京字节跳动科技有限公司 一种全景视频关键帧编码方法和装置
CN106447585A (zh) * 2016-09-21 2017-02-22 武汉大学 城市地区和室内高精度视觉定位系统及方法
CN106767850B (zh) * 2016-11-10 2022-12-30 广州市沃希信息科技有限公司 一种基于场景图片的乘客定位方法及系统
CN107025275B (zh) * 2017-03-21 2019-11-15 腾讯科技（深圳）有限公司 视频搜索方法及装置
CN107027041B (zh) * 2017-04-18 2019-12-27 深圳市智能现实科技有限公司 场景展示方法及装置
CN107146278B (zh) * 2017-04-18 2020-05-26 深圳市智能现实科技有限公司 场景建模方法及装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20010003182A1 (en) * 1999-12-03 2001-06-07 Lilian Labelle Method and devices for indexing and seeking digital images taking into account the definition of regions of interest
CN106909684A (zh) * 2017-03-06 2017-06-30 青岛鹏海软件有限公司 全景图像检索及展示方法
CN108009588A (zh) * 2017-12-01 2018-05-08 深圳市智能现实科技有限公司 定位方法及装置、移动终端

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110852153A (zh) * 2019-09-26 2020-02-28 深圳市金溢科技股份有限公司 一种泊位状态检测方法、获取方法、装置、工控机及介质
CN111314739A (zh) * 2020-02-17 2020-06-19 聚好看科技股份有限公司 一种图像处理方法、服务器及显示设备
CN111582378A (zh) * 2020-05-09 2020-08-25 上海钧正网络科技有限公司 定位识别模型的训练生成方法、位置检测方法、装置
CN111582378B (zh) * 2020-05-09 2024-02-27 上海钧正网络科技有限公司 定位识别模型的训练生成方法、位置检测方法、装置

Also Published As

Publication number Publication date
CN108009588A (zh) 2018-05-08

Similar Documents

Publication Publication Date Title
WO2019104953A1 (fr) Procédé et appareil de localisation et terminal mobile
US11165959B2 (en) Connecting and using building data acquired from mobile devices
US9953506B2 (en) Alarming method and device
CN110674719B (zh) 目标对象匹配方法及装置、电子设备和存储介质
JP6392468B2 (ja) 領域認識方法及び装置
CN108629354B (zh) 目标检测方法及装置
CN111983635B (zh) 位姿确定方法及装置、电子设备和存储介质
US8879803B2 (en) Method, apparatus, and computer program product for image clustering
CN105512685B (zh) 物体识别方法和装置
TW202026948A (zh) 活體檢測方法、裝置以及儲存介質
WO2020019760A1 (fr) Procédé, appareil et système de détection de corps vivant, dispositif électronique et support d'enregistrement
US10382673B2 (en) Automated guide for image capturing for 3D model creation
CN105447864A (zh) 图像的处理方法、装置及终端
CN106557759B (zh) 一种标志牌信息获取方法及装置
CN105160327A (zh) 建筑物识别方法和装置
WO2021031954A1 (fr) Procédé et appareil de détermination de quantité d'objets, ainsi que support de stockage et dispositif électronique
CN109034150B (zh) 图像处理方法及装置
US20210201478A1 (en) Image processing methods, electronic devices, and storage media
CN109684277B (zh) 一种图像显示方法及终端
CN111476057B (zh) 车道线获取方法及装置、车辆驾驶方法及装置
CN111582257A (zh) 用于对待检测对象进行检测的方法、装置及系统
CA3069813C (fr) Capture, connexion et utilisation de donnees d'interieur de batiment a partir de dispositifs mobiles
CN112115894A (zh) 手部关键点检测模型的训练方法、装置及电子设备
CN106980880A (zh) 图像匹配的方法及装置
CN107292901B (zh) 边缘检测方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18883393

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 30/09/2020)

122 Ep: pct application non-entry in european phase

Ref document number: 18883393

Country of ref document: EP

Kind code of ref document: A1