WO2018145246A1 - 一种导盲系统 - Google Patents

一种导盲系统 Download PDF

Info

Publication number
WO2018145246A1
WO2018145246A1 PCT/CN2017/073055 CN2017073055W WO2018145246A1 WO 2018145246 A1 WO2018145246 A1 WO 2018145246A1 CN 2017073055 W CN2017073055 W CN 2017073055W WO 2018145246 A1 WO2018145246 A1 WO 2018145246A1
Authority
WO
WIPO (PCT)
Prior art keywords
coin
image
detected
category
processing module
Prior art date
Application number
PCT/CN2017/073055
Other languages
English (en)
French (fr)
Inventor
刘兆祥
廉士国
Original Assignee
深圳前海达闼云端智能科技有限公司
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 深圳前海达闼云端智能科技有限公司 filed Critical 深圳前海达闼云端智能科技有限公司
Priority to CN201780000652.7A priority Critical patent/CN107278318B/zh
Priority to PCT/CN2017/073055 priority patent/WO2018145246A1/zh
Publication of WO2018145246A1 publication Critical patent/WO2018145246A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00—Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/20—Testing patterns thereon
    • G07D7/2016—Testing patterns thereon using feature extraction, e.g. segmentation, edge detection or Hough-transformation

Definitions

  • the present application relates to the field of guiding blind technology, and in particular to a guiding blind system.
  • Blindness is one of the serious social and public health problems in the world. More than 70% of human information is obtained through vision, and vision problems largely limit access to information for blind people. Due to the defects in the blind eyesight, in the face of a wide variety of coins in the market, the information of the coins cannot be obtained, thus greatly restricting the independent living of the blind.
  • Embodiments of the present application provide a guide blind system for assisting a blind person to obtain coin information.
  • Providing a blind guiding system for assisting a blind person to acquire coin information including: an image capturing device, a processing module, and a voice output device; the image capturing device and the voice output device are coupled to the processing module;
  • the processing module is configured to control the image collection device to acquire a coin image to be detected, acquire a category of the coin according to the to-be-detected coin image, and acquire corresponding coin information according to the category of the coin; the processing module further And configured to control the voice output device to broadcast the coin information by voice.
  • the guiding system includes: an image capturing device, a processing module, and a voice output device; the image capturing device and the voice output device are coupled to the processing module, wherein the processing module can control the image capturing device to obtain the to-be-detected The coin image, the category of the coin is obtained according to the image of the coin to be detected, and the corresponding coin information is obtained according to the category of the coin; the processing module is further configured to control the voice output device to voice through the coin The information is broadcasted.
  • the guide blind system provided by the embodiment of the present application acquires the image of the coin to be detected during image acquisition, there is no need to limit the placement of the coin too much, and the influence of the light intensity and the position of the light when the image is collected is not considered. Therefore, the operation in the process of recognizing the coin is simple and easy, and is suitable for obtaining the coin information in the daily life of the blind person, that is, the embodiment of the present application can assist the blind person to obtain the coin information.
  • FIG. 1 is a schematic structural diagram of a guide blind system provided by an embodiment of the present application.
  • FIG. 2 is a flow chart of steps of a method performed by a blind guiding system according to an embodiment of the present application
  • FIG. 3 is a schematic structural diagram of a detection area provided by an embodiment of the present application.
  • FIG. 5 is a third flowchart of the steps of the method performed by the guide blind system provided by the embodiment of the present application.
  • Figure 6 is a schematic structural diagram of a detection area provided by an embodiment of the present application.
  • Figure 7 is a fourth flowchart of the steps of the method performed by the guide blind system provided by the embodiment of the present application.
  • Figure 8 is a schematic structural diagram of a guide blind system according to an embodiment of the present application.
  • FIG. 9 is a third schematic structural diagram of a guide blind system according to an embodiment of the present application.
  • FIG. 10 is a schematic structural diagram of a guide blind system according to an embodiment of the present application four.
  • the term “and/or” is merely an association relationship describing an association object, indicating that there may be three relationships, for example, A and/or B, which may indicate that A exists separately, and A and B, there are three cases of B alone.
  • the character "/" in this article generally indicates that the contextual object is an "or” relationship. Unless otherwise stated, "multiple" in this document refers to two or more.
  • the words “exemplary” or “such as” are used to mean an example, illustration, or illustration. Any embodiment or design described as “exemplary” or “for example” in the embodiments of the present application should not be construed as preferred or advantageous over other embodiments or designs. Rather, the use of the words “exemplary” or “such as” is intended to present the concepts in a particular manner.
  • the basic principle of the technical solution provided by the embodiment of the present application is that the coin identification method in the prior art is not applicable to the problem of obtaining coin information in the daily life of the blind.
  • the image of the coin is obtained, and then according to the coin.
  • Image of getting money The category of the coin, and then obtain the corresponding coin information according to the coin type; finally, the coin information is broadcast by voice. Since the blind person can obtain the coin information through hearing, and the operation in the process of recognizing the money is simple and easy, the embodiment of the present application can be applied to assist the blind person to obtain the coin information.
  • the embodiment of the present application provides a blind guiding system.
  • the guiding system in the embodiment of the present application may include: a head-mounted blind guiding device, a guiding blind glasses, a mobile phone, a portable computer, a pocket computer, a handheld computer, a digital photo frame, a palmtop computer, a navigation device, and the like.
  • the device, or the guide blind system may also include: a terminal device, a wireless communication network, and a network server.
  • the blind guiding system 10 provided by the embodiment of the present application includes: an image capturing device 11 , a processing module 12 , and a voice output device 13 ; the image capturing device 11 and the voice output device 12 are coupled to the processing module 12 .
  • the processing module 12 is configured to control the image capturing device 11 to acquire the coin image to be detected, obtain the category of the coin according to the coin image to be detected, and obtain the corresponding coin information according to the category of the coin; the processing module 11 is further configured to control the voice output device 13 to pass the voice. Broadcast the coin information.
  • the image capturing module 11 described above may be one or more of an image sensor such as a monocular camera or a binocular camera.
  • the processing module 12 can be a processor or a collective term for multiple processing elements.
  • the central processing unit (English name: central processing unit, referred to as: CPU).
  • the processing module 12 can also be other general purpose processors, digital signal processors (English name: digital signal processing, DSP for short), application specific integrated circuits, field programmable gate arrays or other programmable logic devices, discrete gates or transistor logic.
  • the devices, discrete hardware components, and the like can implement or perform various illustrative logical blocks, modules, and circuits described in connection with the present disclosure.
  • the general purpose processor may be a microprocessor or the processor or any conventional processor or the like.
  • the processing module 12 can also be a dedicated processor, which can include at least one of a baseband processing chip, a radio frequency processing chip, and the like.
  • the processor can also be a combination of computing functions, for example, including one or more microprocessor combinations, a combination of a DSP and a microprocessor, and the like.
  • the voice output device 13 can be Yang Audio output devices such as sounders, amplifiers, speakers, and headphones.
  • the coin in the embodiment of the present application may be a banknote, a coin, or other types of coins.
  • the currency of the coin in the embodiment of the present application may be RMB, US dollars, British pounds, Euros, and the like.
  • the coin information information may include coins: currency (for example: RMB, US dollar, British pound, Euro, etc.), denomination (for example: 1 yuan, 2 yuan, 5 yuan, 10 yuan, 100 yuan, etc.), Version (for example: the fifth set of RMB, the third set of RMB, the new version of the US dollar, etc.) and other information.
  • currency for example: RMB, US dollar, British pound, Euro, etc.
  • denomination for example: 1 yuan, 2 yuan, 5 yuan, 10 yuan, 100 yuan, etc.
  • Version for example: the fifth set of RMB, the third set of RMB, the new version of the US dollar, etc.
  • the processing module 12 controls the image capturing device 11 to acquire the image of the coin to be detected: the processing module can control the image collecting device to perform image capturing on the coin from a viewing angle, or control the image capturing device from multiple perspectives. Coins for image collection. In addition, the same viewing angle can also perform multiple image acquisition on the coin, and then select one or more of the coins as the image to be detected based on the sharpness, completeness, and the like of the captured image.
  • the processing module 12 and the image collection device 11 can be implemented by performing the following operations:
  • the user places the coin within the collection range of the image capture device, then moves and/or flips the coin, and the processing module 12 controls the image capture device 11 to repeatedly press the coin during the user's movement and/or flipping of the coin.
  • Image acquisition is performed to achieve image collection of coins from multiple perspectives.
  • the image capturing device 11 is provided to include a plurality of devices that can independently acquire images, and a plurality of devices that can independently acquire images are respectively disposed at different positions.
  • the processing module 12 controls a plurality of image collection devices disposed at different positions to perform image collection on the coins from one perspective, thereby performing image collection on the coins from a plurality of perspectives.
  • the image acquisition device includes two opposite cameras. When the image is collected and collected by the coin, the user places the coin between the two cameras, and the two cameras respectively collect images of the front and back of the coin.
  • the guiding system includes: an image collecting device and a processing module The group and the voice output device; the image capturing device and the voice output device are coupled to the processing module, wherein the processing module can control the image capturing device to obtain the image of the coin to be detected, obtain the category of the coin according to the image of the coin to be detected, and obtain the category according to the coin Corresponding coin information; the processing module is further configured to control the voice output device to broadcast the coin information by voice, and the guide blind system provided by the embodiment of the present application does not need to place the coin placement position when performing image acquisition to obtain the coin image to be detected.
  • the limitation is imposed, and the influence of the illumination intensity, the illumination position, and the like in the image collection is not considered, so the operation in the process of recognizing the coin is simple and easy, and is suitable for obtaining the coin information in the daily life of the blind, that is, the embodiment of the present application can assist the blind person Get coin information.
  • the processing module 12 acquires the category of the coin according to the to-be-detected coin image and acquires the corresponding coin information according to the category of the coin, including: the processing module 12 intercepts the detection area on the image of the coin to be detected; and detects the coin to be detected.
  • the center of the image is centered; processing the image of the intercepted detection area to obtain a second processed image; the second processed image is a grayscale image and the pixel size of the second processed image is a preset value; and the second processed image is extracted from the second processed image Detecting features; respectively calculating the Euclidean distance between the detection feature and the feature template of each coin (English name: Euclidean Metric, nickname: Euclidean metric); obtaining matching feature template; matching feature template is the smallest Euclidean distance from the detected feature The feature template; obtaining the category of the coin according to the matching feature template and obtaining the corresponding coin information according to the category of the coin.
  • the processing module 12 can implement the following steps to obtain the category of the coin according to the coin image to be detected and the corresponding coin information according to the category of the coin.
  • the detection area is intercepted on the image of the coin to be detected.
  • the detection area is centered on the center of the coin image to be detected.
  • the center of the coin image 31 to be detected is 310, and the center of the detection area 32 coincides with the center 310 of the coin image 31 to be detected.
  • the coin image 31 to be detected and the detection area 32 are both rectangular.
  • the embodiment of the present application is not limited thereto, and the coin image 31 and the detection area 32 to be detected may also be other. shape.
  • the second processed image is a grayscale image and the pixel size of the second processed image is a preset value.
  • processing the image of the intercepted detection area to obtain the second processed image may be: first converting the image of the detection area into a grayscale image, and then normalizing the grayscale image.
  • the image of the detection area can be processed by other image processing methods.
  • the image processing of the image of the detection area is not limited in the embodiment of the present application, so that the image of the detection area can be processed into a gray of a preset pixel size.
  • the order image is subject to change.
  • the preset value may specifically be 100 pixels * 100 pixels.
  • the detection feature may be: a scale-invariant feature transform (SIFT) feature.
  • SIFT scale-invariant feature transform
  • the matching feature template is a feature template with the smallest Euclidean distance from the detected feature.
  • the processing module 12 needs to calculate the Euclidean distance between the detection feature and the template feature of each coin, so before the step S24, the processing module 12 needs to extract and save the feature template of each coin so that The feature template of each coin is read in step S24 for calculation. Therefore, the processing module 12 is further configured to acquire a template image of each coin; and process the template image of each coin to obtain a corresponding first processed image; the first processed image is a grayscale image and the pixel size of the first processed image is a preset Value; feature templates are extracted from each of the first processed images and saved.
  • the processing module 12 extracts and saves the feature template of each coin by performing the following steps.
  • the template image for obtaining each coin may specifically be a template image of obtaining a manuscript image of a front side and a reverse side of various coins as a coin.
  • the template image for obtaining the US dollar may be: obtaining a draft image of $1, $2, $5, $10, $20, $50, and $100 for the front and back original scales, respectively, and obtaining 14 images.
  • the first processed image is a grayscale image and the pixel size of the first processed image is a preset value.
  • processing the template image of each coin to obtain the first processed image may be: first converting the template image of each coin into a grayscale image, and then normalizing the grayscale image.
  • the image processing of the template image of each coin is not limited in the embodiment of the present application, so that the template image of each coin can be processed into a grayscale image whose pixel size is a preset value.
  • the processing module 12 may save the template features of each coin in a local database, or may save the template features of each coin in the remote server.
  • the detection feature and the template feature are the same type of feature.
  • the detection feature is also a SIFT feature.
  • steps S41, S42 and S43 are feature template extraction and save processes, and in some embodiments, it is not necessary to perform the above S41, S42 and S43.
  • the template feature has been extracted and saved in the local database when it is first used.
  • the template feature can be retrieved directly from the local database without repeating the above steps. S41, S42 and S43.
  • the template features can be enriched, and then each type of coin can be identified and classified when acquiring the category of the coin, but currently there are 1200 in the world.
  • the processing module 12 is further configured to determine whether a minimum value of the Euclidean distance between the detection feature and each feature template is less than or equal to a threshold; if so, a feature template having the smallest Euclidean distance of the detected feature is used as a matching feature template; No, determining whether the number of times the image of the coin to be detected is rotated is equal to a preset number; if the number of times the image of the coin to be detected has been rotated is less than a preset number of times, the image of the coin to be detected is rotated by a preset angle and the detection area is re-intercepted;
  • the processing module 12 is configured to intercept the detection area on the image of the coin to be detected, including: intercepting the detection area on the image of the coin to be detected after acquiring the image of the coin to be detected, and after rotating the image of the coin to be detected, on the image of the coin to be detected Intercept the detection area.
  • the processing module 12 acquires the matching feature template by performing the following steps.
  • step S281 if the minimum value of the Euclidean distance between the detected feature and each feature template is less than or equal to the threshold, step S282 is performed; if the minimum value of the Euclidean distance of the detected feature and each feature template is greater than the threshold, step S283 is performed. .
  • a feature template that minimizes an Euclidean distance of the detected feature is used as a matching feature template.
  • the preset number of times may be 1, 2, 3, 10, and the like.
  • step S283 if the number of times the coin image to be detected has been rotated is less than a preset number of times, for example, the preset number of times is 1 time, the image of the coin to be detected has not been rotated; for example, the preset number of times is 10 times, and the coin is to be detected.
  • the image is rotated 5 times; then step S284 is performed and the process returns to step S21 to re-execute steps S21 to S283; if the number of times the coin image to be detected has been rotated is equal to the preset number of times, it indicates that the detection feature is not successfully matched with the template feature. End the test. At this time, "undetected coins" or "detection failure" can be output.
  • the angle a at which the image of the coin to be detected is rotated is 45 degrees is illustrated in FIG. 6 .
  • the preset angle may also be other angles, for example, 10 degrees, 30 degrees, 90 degrees, and the like.
  • the rotation preset angle may be a clockwise rotation of the preset angle or a counterclockwise rotation of the preset angle.
  • the preset number of times can be set according to actual needs.
  • the processing module 12 obtains the category of the coin according to the to-be-detected coin image and acquires the corresponding coin information according to the category of the coin, including: the processing module extracts the preset feature in the to-be-detected coin image; and classifies the preset feature. Obtain the category of the coin and obtain the corresponding coin information according to the category of the coin.
  • the processing module 12 can also obtain the category of the coin and obtain the corresponding coin information according to the category of the coin by performing the following steps.
  • the preset feature in the foregoing embodiment may be a manually set feature, for example.
  • a manually set feature for example.
  • Haar features English name: Haar-like features
  • direction gradient histogram English name: Histogram of Oriented Gradient, referred to as: HOG
  • HOG direction gradient histogram
  • LBP Local Binary Pattern
  • the preset feature in the above embodiment may also be an automatic learning of the acquired image feature by a machine learning method, such as by a deep learning method.
  • the classification method for classifying the preset features in the foregoing embodiment may be: using a support vector machine (English name: Support Vector Machine, SVM for classification), classifying (it name: adaboost) classifier for classification or The nearest neighbor algorithm classifies, and then obtains corresponding coin information according to the category of the coin.
  • a support vector machine English name: Support Vector Machine, SVM for classification
  • classifying it name: adaboost
  • adaboost adaboost classifier for classification
  • the nearest neighbor algorithm classifies, and then obtains corresponding coin information according to the category of the coin.
  • the processing module 12 obtains the category of the coin according to the to-be-detected coin image and acquires the corresponding coin information according to the category of the coin, including: the processing module is specifically configured to use the deep neural network classification method to classify the coin image to obtain the coin.
  • the category and the corresponding coin information are obtained according to the category of the coin.
  • the detected coin image is directly classified by the pre-trained deep neural network classifier, thereby acquiring the category of the coin and acquiring the corresponding coin information according to the category of the coin.
  • the processing module 12 includes: a first processing unit 121, a second processing unit 122, and a network communication unit 123;
  • the first processing unit 121 is configured to control the image collecting device to acquire the image of the coin to be detected, and send the collected image to the second processing unit 122 through the network communication unit 123;
  • the second processing unit 122 is configured to acquire the category of the coin according to the to-be-detected coin image and acquire the corresponding coin information according to the category of the coin, and return the acquired coin information to the first processing unit 121 through the network communication unit 123.
  • the above-mentioned implementation process of acquiring the coin according to the coin image to be detected and acquiring the corresponding coin information according to the category of the coin may be completed inside the terminal device or assisted by the remote service device.
  • the network communication unit 123 of the processing module 12 acquires the coin to be detected acquired by the first processing unit 121.
  • the image is sent to the second processing unit 122.
  • the second processing unit 122 acquires the category of the coin according to the coin image to be detected and acquires the corresponding coin information according to the category of the coin, and returns the acquired coin information to the first through the network communication unit 123.
  • a processing unit 123 A processing unit 123.
  • the guiding system further includes a helmet 900 ; the image capturing device 11 , the processing module 12 , and the voice output device 13 are disposed on the helmet 900 .
  • the guide blind system further includes: a voice input device 14;
  • the processing module 12 is further configured to collect a voice instruction by using the voice input device 14; specifically, when acquiring an instruction for indicating coin identification, the image capturing device 11 is controlled to acquire a coin image to be detected, and the coin is obtained according to the image of the coin to be detected.
  • the category and the corresponding coin information are obtained according to the category of the coin. In this way, the guide blind system can enter the flow of the coin image acquisition according to the trigger of the blind person, thereby adopting the template of the corresponding coin recognition, which can effectively improve the efficiency and precision of the coin recognition.
  • the voice input device 14 can be a microphone.
  • the functions described herein can be implemented in hardware, software, firmware, or any combination thereof.
  • the functions may be stored in a computer readable medium or transmitted as one or more instructions or code on a computer readable medium.
  • Computer readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one location to another.
  • a storage medium may be any available media that can be accessed by a general purpose or special purpose computer.

Abstract

一种导盲系统（10），涉及导盲技术领域，用于辅助盲人获取钱币信息。该导盲系统（10）包括：图像采集设备（11）、处理模组（12）以及语音输出设备（13）；图像采集设备（11）和语音输出设备（13）耦合至处理模组（12）；处理模组（12）用于控制图像采集设备（11）获取待检测钱币图像（31），根据待检测钱币图像（31）获取钱币的类别以及根据钱币的类别获取对应的钱币信息；处理模组（12）还用于控制语音输出设备（13）通过语音对钱币信息进行播报。该导盲系统（10）用于辅助盲人获取钱币信息。

Description

一种导盲系统 技术领域
本申请涉及导盲技术领域，尤其涉及一种导盲系统。
背景技术
盲是世界上严重的社会和公共卫生问题之一。人类70％以上的信息都是通过视觉来获取的，视力问题很大程度上限制了盲人进行信息获取。由于盲人视力上的缺陷，在面对市面种类繁多的钱币时，无法获取钱币的信息，因此很大程度上限制了盲人独立生活。
发明内容
本申请的实施例提供一种导盲系统，用于辅助盲人获取钱币信息。
为达到上述目的，本申请的实施例采用如下技术方案：
提供一种导盲系统，用于辅助盲人获取钱币信息，包括：图像采集设备、处理模组以及语音输出设备；所述图像采集设备和所述语音输出设备耦合至所述处理模组；
所述处理模组用于控制所述图像采集设备获取待检测钱币图像，根据所述待检测钱币图像获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息；所述处理模块还用于控制所述语音输出设备通过语音对所述钱币信息进行播报。
本申请的实施例提供的导盲系统包括：图像采集设备、处理模组以及语音输出设备；图像采集设备和语音输出设备耦合至处理模组，其中，处理模组可以控制图像采集设备获取待检测钱币图像，根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息；处理模块还用于控制语音输出设备通过语音对钱币 信息进行播报，由于本申请实施例提供的导盲系统在进行图像采集获取待检测钱币图像时，无需过多对钱币放置位置进行限制，且不用考虑图像采集时的光照强度、光照位置等的影响，所以对钱币的识别过程中的操作简单易行，适用于盲人日常生活中获取钱币信息，即本申请实施例可以辅助盲人获取钱币信息。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请的实施例提供的导盲系统的示意性结构图之一；
图2为本申请的实施例提供的导盲系统所执行的方法的步骤流程图之一；
图3为本申请的实施例提供的检测区域的示意性结构图之一；
图4为本申请的实施例提供的导盲系统所执行的方法的步骤流程图之二；
图5为本申请的实施例提供的导盲系统所执行的方法的步骤流程图之三；
图6为本申请的实施例提供的检测区域的示意性结构图之二；
图7为本申请的实施例提供的导盲系统所执行的方法的步骤流程图之四；
图8为本申请的实施例提供的导盲系统的示意性结构图之二；
图9为本申请的实施例提供的导盲系统的示意性结构图之三；
图10为本申请的实施例提供的导盲系统的示意性结构图之 四。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。需要说明的是，下文所提供的任意多个技术方案中的部分或全部技术特征在不冲突的情况下，可以结合使用，形成新的技术方案。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
在本申请实施例中术语“和/或”，仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。另外，本文中字符“/”，一般表示前后关联对象是一种“或”的关系。如果不加说明，本文中的“多个”是指两个或两个以上。
在本申请实施例中，“示例性的”或者“例如”等词用于表示作例子、例证或说明。本申请实施例中被描述为“示例性的”或者“例如”的任何实施例或设计方案不应被解释为比其它实施例或设计方案更优选或更具优势。确切而言，使用“示例性的”或者“例如”等词旨在以具体方式呈现相关概念。
需要说明的是，本申请实施例中，除非另有说明，“多个”的含义是指两个或两个以上。
还需要说明的是，本申请实施例中，“的(英文：of)”，“相应的(英文：corresponding，relevant)”和“对应的(英文：corresponding)”有时可以混用，应当指出的是，在不强调其区别时，其所要表达的含义是一致的。
本申请实施例所提供的技术方案的基本原理为：针对现有技术中的钱币识别方式不适用于盲人日常生活中获取钱币信息的问题，本申请实施例中通过获取钱币的图像，然后根据钱币的图像获取钱 币的类别，再根据钱币类别获取对应的钱币信息；最后通过语音对钱币信息进行播报。由于盲人可以通过听觉获取钱币信息，而且对钱币的识别过程中的操作简单易行，所以本申请实施例可以适用于辅助盲人获取钱币信息。
基于上述内容，本申请实施例提供一种导盲系统。
示例性的，本申请实施例中的导盲系统可以包括：头戴式导盲装置、导盲眼镜、手机、便携式计算机、袖珍式计算机、手持式计算机、数码相框、掌上电脑、导航仪等终端设备，或者导盲系统也可以包括：终端设备、无线通信网络以及网络服务器。
参照图1所示，本申请实施例提供的导盲系统10包括：图像采集设备11、处理模组12以及语音输出设备13；图像采集设备11和语音输出设备12耦合至处理模组12。
处理模组12用于控制图像采集设备11获取待检测钱币图像，根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息；处理模块11还用于控制语音输出设备13通过语音对钱币信息进行播报。
示例性的，上述的图像采集模块11可以是单目摄像头、双目摄像头等图像传感器中的一种或多种。处理模组12可以是处理器，也可以是多个处理元件的统称。例如，中央处理器(英文名称：central processing unit，简称：CPU)。处理模组12还可以为其他通用处理器、数字信号处理器(英文名称：digital signal processing，简称：DSP)、专用集成电路、现场可编程门阵列或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等，其可以实现或执行结合本申请公开内容所描述的各种示例性的逻辑方框，模块和电路。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。处理模组12还可以为专用处理器，该专用处理器可以包括基带处理芯片、射频处理芯片等中的至少一个。处理器也可以是实现计算功能的组合，例如包含一个或多个微处理器组合，DSP和微处理器的组合等。语音输出设备13可以是扬 声器、功放机、音箱、耳机等音频输出设备。
需要说明的是，本申请实施例中的钱币可以为纸币，也可以为硬币，还可以其他类型的钱币。此外，本申请实施例中的钱币的币种可以为人民币、美元、英镑、欧元等。
还需要说明的是，钱币信息信息可以包括钱币的：币种(例如：人民币、美元、英镑、欧元等)、面额(例如：1元、2元、5元、10元、100元等)、版本(例如：第五套人民币、第三套人民币、新版美元等)等信息。
示例性的，在处理模组12控制图像采集设备11获取待检测钱币图像的过程中：处理模组可以控制图像采集设备从一个视角对钱币进行图像采集，或者控制图像采集设备从多个视角对钱币进行图像采集。此外，同一视角也可以对钱币进行多次图像采集，然后基于采集的图像的清晰度、完整度等选取其中的一张或多张作为待检测钱币图像。
此外，当从多个视角对钱币进行图像采集时，处理模组12和图像采集设备11可以通过执行如下操作来实现：
用户将钱币放置于图像采集装置的采集范围之内，然后对钱币进行移动和/或翻转，处理模组12控制图像采集装置11在用户对钱币进行移动和/或翻转的过程中多次对钱币进行图像采集，从而实现从多个视角对钱币进行图像采集。或者设置图像采集设备11包括多个可以独立采集图像的装置，且多个可以独立采集图像的装置分别设置于不同的位置。当进行图像采集时，处理模组12控制多个设置于不同位置的图像采集装置分别从一个视角对钱币进行图像采集，从而实现从多个视角对钱币进行图像采集。例如：图像采集设备包括两个相对的摄像头，对钱币进行图像采集采集时，用户将钱币放置在两个摄像头之间，两个摄像头分别采集钱币正面和反面的图像。
本申请的实施例提供的导盲系统包括：图像采集设备、处理模 组以及语音输出设备；图像采集设备和语音输出设备耦合至处理模组，其中，处理模组可以控制图像采集设备获取待检测钱币图像，根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息；处理模块还用于控制语音输出设备通过语音对钱币信息进行播报，由于本申请实施例提供的导盲系统在进行图像采集获取待检测钱币图像时，无需过多对钱币放置位置进行限制，且不用考虑图像采集时的光照强度、光照位置等的影响，所以对钱币的识别过程中的操作简单易行，适用于盲人日常生活中获取钱币信息，即本申请实施例可以辅助盲人获取钱币信息。
可选的，处理模组12根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息，包括：处理模组12在待检测钱币图像上截取检测区域；检测区域以待检测钱币图像的中心为中心；对截取到的检测区域的图像进行处理获取第二处理图像；第二处理图像为灰阶图像且第二处理图像的像素大小为预设值；从第二处理图像中提取检测特征；分别计算检测特征与各个钱币的特征模板的欧氏距离(英文名称：Euclidean Metric，别称：欧几里得度量)；获取匹配特征模板；匹配特征模板为与检测特征的欧氏距离最小的特征模板；根据匹配特征模板获取钱币的类别以及根据钱币的类别获取对应的钱币信息。
即，参照图2所示，处理模组12可以通过执行如下步骤来实现根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息。
S21、在待检测钱币图像上截取检测区域。
其中，检测区域以待检测钱币图像的中心为中心。
示例性的，参照图3所示，待检测钱币图像31的中心为310，检测区域32的中心与待检测钱币图像31的中心310重合。需要说明的是，图3中以待检测钱币图像31和检测区域32均为矩形为例进行说明，但本申请实施例并不限定于此，待检测钱币图像31和检测区域32也可以为其他形状。
S22、对截取到的检测区域的图像进行处理获取第二处理图像。
其中，第二处理图像为灰阶图像且第二处理图像的像素大小为预设值。
示例性的，对截取到的检测区域的图像进行处理获取第二处理图像具体可为：首先将检测区域的图像转换为灰阶图像，然后对灰阶图像进行归一化处理。当然也可以通过其他图像处理方式检测区域的图像进行处理，本申请实施例中不限定对检测区域的图像进行图像处理的方式，以能够将检测区域的图像处理为像素大小为预设值的灰阶图像为准。
示例性的，预设值具体可以为100像素*100像素。
S23、从第二处理图像中提取检测特征。
示例性的，检测特征具体可以为：尺寸尺度不变特征变换(英文名称：Scale-invariant Feature Transform，简称：SIFT)特征。
S24、分别计算检测特征与各个钱币的模板特征的欧氏距离。
S25、获取匹配特征模板。
其中，匹配特征模板为与检测特征的欧氏距离最小的特征模板。
S26、根据匹配特征模板获取钱币的类别以及根据钱币的类别获取对应的钱币信息。
进一步的，在上述步骤S24中处理模组12需要计算检测特征与各个钱币的模板特征的欧氏距离，因此在步骤S24之前处理模组12还需要对各个钱币的特征模板进行提取并保存，以便于在步骤S24中读取各个钱币的特征模板进行计算。因此处理模组12还用于获取各个钱币的模板图像；对各个钱币的模板图像进行处理获取对应的第一处理图像；第一处理图像为灰阶图像且第一处理图像的像素大小为预设值；从各个第一处理图像中提取特征模板并保存。
即，参照图4所示，处理模组12对各个钱币的特征模板进行进行提取并保存可以通过执行如下步骤来实现。
S41、获取各个钱币的模板图像。
具体的，获取各个钱币的模板图像具体可以为获取各种钱币正面、反面原始比例的稿清图像作为钱币的模板图像。例如：获取美元的模板图像具体可以为：分别获取1美元、2美元、5美元、10美元、20美元、50美元以及100美元正面、反面原始比例的稿清图像，并将获取的14张图像作为钱币的模板图像。
S42、对各个钱币的模板图像进行处理获取对应的第一处理图像。
其中，第一处理图像为灰阶图像且第一处理图像的像素大小为预设值。
示例性的，对各个钱币的模板图像进行处理获取第一处理图像具体可为：首先将各个钱币的模板图像转换为灰阶图像，然后对灰阶图像进行归一化处理。同样，本申请实施例中也不限定对各个钱币的模板图像进行图像处理的方式，以能够将各个钱币的模板图像处理为像素大小为预设值的灰阶图像为准。
S43、从各个第一处理图像中提取模板特征并保存。
示例性的，处理模组12可以将各个钱币的模板特征保存在本地数据库中，也可以将各个钱币的模板特征保存在远端服务器中。
需要说明的是，检测特征与模板特征为相同类型的特征。例如：模板特征为SIFT特征，则检测特征也为SIFT特征。
还需要说明的是，上述步骤S41、S42和S43为特征模板提取及保存过程，在一些实施例中，可以无需执行上述S41、S42和S43。例如：在首次使用时已将模板特征提取并保存在本地数据库中，在下一次通过图2所示实现方式来获取钱币信息时，可以直接从本地数据库中调取模板特征，而无需重复执行上述步骤S41、S42和S43。
还需要说明的是，若对每一种钱币均执行上述步骤S41、S42和S43可以丰富模板特征，进而可以在获取钱币的类别时可以对每一种钱币进行识别分类，但目前世界已存在1200多种钱币，钱币种类繁多，若对每一个钱币均执行上述步骤S41、S42和S43，则需要较多的计算资源、存储资源，而大部分模板特征在盲人日常生活中用不到，造成资源浪费，因此在实际使用时可以仅对一部分钱币执行上述步骤S41、S42和S43，且用户可以根据实际需求增加或删除部分模板特征。
进一步的，处理模组12还用于判断检测特征与各个特征模板的欧氏距离的最小值是否小于或等于阈值；若是，将与检测特征的欧氏距离最小的特征模板作为匹配特征模板；若否，判断已对待检测钱币图像旋转的次数是否等于预设次数；若已对待检测钱币图像旋转的次数小于预设次数，对待检测钱币图像旋转预设角度并重新截取检测区域；
处理模组12用于在待检测钱币图像上截取检测区域，包括：在获取到待检测钱币图像后在待检测钱币图像上截取检测区域以及在旋转待检测钱币图像后，在待检测钱币图像上截取检测区域。
即，参照图5所示，上述步骤S25中，处理模组12获取匹配特征模板可以通过执行如下步骤来实现。
S281、判断检测特征与各个特征模板的欧氏距离的最小值是否小于或等于阈值。
在步骤S281中，若检测特征与各个特征模板的欧氏距离的最小值小于或等于阈值，则执行步骤S282；若检测特征与各个特征模板的欧氏距离的最小值大于阈值，则执行步骤S283。
S282、将与检测特征的欧氏距离最小的特征模板作为匹配特征模板。
S283、判断已对待检测钱币图像旋转的次数是否等于预设次数。
示例性的，预设次数可以为1次、2次、3次、10次等。
在步骤S283中，若已对待检测钱币图像旋转的次数小于预设次数，例如：预设次数为1次，尚未对待检测钱币图像进行过旋转；再例如：预设次数为10次，对待检测钱币图像进行过5次旋转；则执行步骤S284并返回步骤S21中重新执行步骤S21至S283；若已对待检测钱币图像旋转的次数等于预设次数，则说明未能将检测特征与模板特征匹配成功，结束检测。此时可以输出“未检测到钱币”或者“检测失败”。
S284、将待检测钱币图像旋转预设角度。
示例性的，参照图6所示，图6中以将待检测钱币图像旋转的角度a为45度为例进行说明。需要说明的是，预设角度还可以为其他角度，例如：10度、30度、90度等。此外，旋转预设角度可以为顺时针旋转预设角度，也可以为逆时针旋转预设角度。
需要说明的是，在上述实施例中，若预设次数设置的越大，则步骤S21至S283被重复执行的次数越多，计算耗时越长，但将检测特征与模板特征匹配成功的可能性越大。反之，若预设次数设置的越小，步骤S21至S283被重复执行的次数越少，计算耗时越短，但将检测特征与模板特征匹配成功的可能性越小。因此预设次数的大小可以根据实际需求进行设定。
可选的，处理模组12根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息，包括：处理模组在待检测钱币图像中提取预设特征；对预设特征进行分类获取钱币的类别以及根据钱币的类别获取对应的钱币信息。
即，参照图7所示，处理模组12还可以通过执行如下步骤来获取钱币的类别以及根据钱币的类别获取对应的钱币信息。
S71、在待检测钱币图像中提取预设特征。
可选的，上述实施例中的预设特征可以是手动设置的特征，例 如：Haar特征(英文名称：Haar-like features)、方向梯度直方图(英文名称：Histogram of Oriented Gradient，简称：HOG)特征、局部二值模式(英文名称：Local Binary Pattern，简称：LBP)特征。上述实施例中的预设特征也可以是通过机器学习方法，比如通过深度学习方法自动学习获取的图像特征。
S72、对预设特征进行分类获取钱币的类别以及根据钱币的类别获取对应的钱币信息。
具体的，上述实施例中对预设特征进行分类的分类方法可以为：采用支持向量机(英文名称：Support Vector Machine，简称：SVM)进行分类、迭代(英文名称：adaboost)分类器进行分类或者最邻近算法进行分类，进而根据钱币的类别获取对应的钱币信息。
可选的，处理模组12根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息，包括：处理模组具体用于通过深度神经网络分类法对待检测钱币图像进行分类获取钱币的类别以及根据钱币的类别获取对应的钱币信息。
即，直接通过预先训练好的深度神经网络分类器对待检测钱币图像进行分类，从而获取钱币的类别并根据钱币的类别获取对应的钱币信息。
可选的，参照图8所示，处理模组12包括：第一处理单元121、第二处理单元122以及网络通信单元123；
第一处理单元121，用于控制图像采集设备获取待检测钱币图像，并将采集到的图像通过网络通信单元123发送至第二处理单元122；
第二处理单元122用于根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息，并通过网络通信单元123将获取到的钱币信息返回至第一处理单元121。
即，上述根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息的实现过程可以在终端设备内部完成，也可以通过远程服务设备协助完成。当根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息的过程通过远程服务设备协助完成时，处理模组12的网络通信单元123将第一处理单元121获取的待检测钱币图像发送至第二处理单元122中，第二处理单元122根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息，并通过网络通信单元123将获取到的钱币信息返回至第一处理单元123。
可选的，参照图9所示，导盲系统还包括头盔900；图像采集设备11、处理模组12以及语音输出设备13设置在头盔900上。
可选的，参照图10所示，导盲系统还包括：语音输入设备14；
处理模组12还用于通过语音输入设备14采集语音指令；具体用于当采集到用于指示进行钱币识别的指令时，控制图像采集设备11获取待检测钱币图像，根据待检测钱币图像获取钱币的类别以及根据钱币的类别获取对应的钱币信息。通过这种方式，能够使得导盲系统根据盲人的触发进入到钱币图像获取的流程，从而采用相应的钱币识别的模板进行识别，这样能够有效的提高钱币识别的效率和精度。
示例性的，语音输入设备14可以为麦克风。
本领域技术人员应该可以意识到，在上述一个或多个示例中，本申请所描述的功能可以用硬件、软件、固件或它们的任意组合来实现。当使用软件实现时，可以将这些功能存储在计算机可读介质中或者作为计算机可读介质上的一个或多个指令或代码进行传输。计算机可读介质包括计算机存储介质和通信介质，其中通信介质包括便于从一个地方向另一个地方传送计算机程序的任何介质。存储介质可以是通用或专用计算机能够存取的任何可用介质。
以上所述，仅为本申请的具体实施方式，但本申请的保护范围 并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到的变化或替换，都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以权利要求的保护范围为准。

Claims (9)

  1. 一种导盲系统，其特征在于，包括：图像采集设备、处理模组以及语音输出设备；所述图像采集设备和所述语音输出设备耦合至所述处理模组；
    所述处理模组用于控制所述图像采集设备获取待检测钱币图像，根据所述待检测钱币图像获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息；所述处理模块还用于控制所述语音输出设备通过语音对所述钱币信息进行播报。
  2. 根据权利要求1所述的系统，其特征在于，所述处理模组根据所述待检测钱币图像获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息，包括：所述处理模组在所述待检测钱币图像上截取检测区域；所述检测区域以所述待检测钱币图像的中心为中心；对截取到的检测区域的图像进行处理获取第二处理图像；所述第二处理图像为灰阶图像且所述第二处理图像的像素大小为预设值；从所述第二处理图像中提取检测特征；分别计算所述检测特征与各个钱币的特征模板的欧氏距离；获取匹配特征模板；所述匹配特征模板为与所述检测特征的欧氏距离最小的特征模板；根据所述匹配特征模板获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息。
  3. 根据权利要求2所述的系统，其特征在于，所述处理模组还用于获取各个钱币的模板图像；对各个钱币的模板图像进行处理获取对应的第一处理图像；所述第一处理图像为灰阶图像且所述第一处理图像的像素大小为预设值；从各个第一处理图像中提取特征模板并保存。
  4. 根据权利要求2所述的系统，其特征在于，所述处理模组还用于判断所述检测特征与各个钱币的特征模板的欧氏距离的最小值是否小于或等于阈值；若是，将与所述检测特征的欧氏距离最小的特征模板作为所述匹配特征模板；若否，判断已对所述待检测钱币图像旋转的次数是否等于预设次数；若已对所述待检测钱币图像旋 转的次数小于预设次数，则对所述待检测钱币图像旋转预设角度并重新截取检测区域；
    所述处理模组用于在所述待检测钱币图像上截取检测区域，包括：在获取到待检测钱币图像后在所述待检测钱币图像上截取检测区域以及在旋转待检测钱币图像后，在所述待检测钱币图像上截取检测区域。
  5. 根据权利要求1所述的系统，其特征在于，所述处理模组根据所述待检测钱币图像获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息，包括：所述处理模组在所述待检测钱币图像中提取预设特征；对所述预设特征进行分类获取所述钱币的类别，根据所述钱币的类别获取对应的钱币信息。
  6. 根据权利要求1所述的系统，其特征在于，所述处理模组根据所述待检测钱币图像获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息，包括：所述处理模组通过深度神经网络分类法对所述待检测钱币图像进行分类获取所述钱币的类别，根据所述钱币的类别获取对应的钱币信息。
  7. 根据权利要求1-6任一项所述的系统，其特征在于，所述处理模组包括：第一处理单元、第二处理单元以及网络通信单元；
    所述第一处理单元，用于控制所述图像采集设备获取待检测钱币图像，并将采集到的图像通过网络通信单元发送至所述第二处理单元；
    所述第二处理单元用于根据所述待检测钱币图像获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息，并通过网络通信单元将获取到的钱币信息返回至所述第一处理单元。
  8. 根据权利要求1所述的系统，其特征在于，所述导盲系统包括头盔；所述图像采集设备、处理模组以及语音输出设备设置在所述头盔上。
  9. 根据权利要求1所述的系统，其特征在于，所述系统还包括： 语音输入设备；
    所述处理模组还用于通过所述语音输入设备采集语音指令；具体用于当采集到用于指示进行钱币识别的指令时，控制所述图像采集设备获取待检测钱币图像，根据所述待检测钱币图像获取所述钱币的类别以及根据所述钱币的类别获取对应的钱币信息。
PCT/CN2017/073055 2017-02-07 2017-02-07 一种导盲系统 WO2018145246A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201780000652.7A CN107278318B (zh) 2017-02-07 2017-02-07 一种导盲系统
PCT/CN2017/073055 WO2018145246A1 (zh) 2017-02-07 2017-02-07 一种导盲系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2017/073055 WO2018145246A1 (zh) 2017-02-07 2017-02-07 一种导盲系统

Publications (1)

Publication Number Publication Date
WO2018145246A1 true WO2018145246A1 (zh) 2018-08-16

Family

ID=60076516

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/073055 WO2018145246A1 (zh) 2017-02-07 2017-02-07 一种导盲系统

Country Status (2)

Country Link
CN (1) CN107278318B (zh)
WO (1) WO2018145246A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI749751B (zh) * 2020-09-04 2021-12-11 南開科技大學 盲人用紙鈔識別系統及其方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6748097B1 (en) * 2000-06-23 2004-06-08 Eastman Kodak Company Method for varying the number, size, and magnification of photographic prints based on image emphasis and appeal
CN101814128A (zh) * 2009-02-20 2010-08-25 窦鹃鹃 一种盲人使用的便携式美元纸币面额识别机
CN204029024U (zh) * 2014-07-04 2014-12-17 宁夏诚洋数字科技有限公司 一种纸币面值识别系统
CN104464079A (zh) * 2014-12-29 2015-03-25 北京邮电大学 基于模板特征点及其拓扑结构的多币种面值识别方法
CN105184225A (zh) * 2015-08-11 2015-12-23 深圳市倍量科技有限公司 一种多国纸币图像识别方法和装置

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010081480A (ja) * 2008-09-29 2010-04-08 Fujifilm Corp 携帯型不審者検出装置、不審者検出方法及びプログラム
CN103048331B (zh) * 2012-12-20 2015-09-23 石家庄印钞有限公司 基于柔性模板配准的印刷缺陷检测方法
CN105787442B (zh) * 2016-02-19 2019-04-30 电子科技大学 一种基于视觉交互面向视障人群的穿戴式辅助系统及其使用方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6748097B1 (en) * 2000-06-23 2004-06-08 Eastman Kodak Company Method for varying the number, size, and magnification of photographic prints based on image emphasis and appeal
CN101814128A (zh) * 2009-02-20 2010-08-25 窦鹃鹃 一种盲人使用的便携式美元纸币面额识别机
CN204029024U (zh) * 2014-07-04 2014-12-17 宁夏诚洋数字科技有限公司 一种纸币面值识别系统
CN104464079A (zh) * 2014-12-29 2015-03-25 北京邮电大学 基于模板特征点及其拓扑结构的多币种面值识别方法
CN105184225A (zh) * 2015-08-11 2015-12-23 深圳市倍量科技有限公司 一种多国纸币图像识别方法和装置

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI749751B (zh) * 2020-09-04 2021-12-11 南開科技大學 盲人用紙鈔識別系統及其方法

Also Published As

Publication number Publication date
CN107278318A (zh) 2017-10-20
CN107278318B (zh) 2019-10-29

Similar Documents

Publication Publication Date Title
JP6844038B2 (ja) 生体検出方法及び装置、電子機器並びに記憶媒体
US10922529B2 (en) Human face authentication method and apparatus, and storage medium
CN109697416B (zh) 一种视频数据处理方法和相关装置
WO2020063744A1 (zh) 人脸检测方法及装置、业务处理方法、终端设备及存储介质
WO2019232866A1 (zh) 人眼模型训练方法、人眼识别方法、装置、设备及介质
Singh et al. Currency recognition on mobile phones
Merino-Gracia et al. A head-mounted device for recognizing text in natural scenes
WO2020252917A1 (zh) 一种模糊人脸图像识别方法、装置、终端设备及介质
WO2019232862A1 (zh) 嘴巴模型训练方法、嘴巴识别方法、装置、设备及介质
TW202006602A (zh) 三維臉部活體檢測方法、臉部認證識別方法及裝置
WO2019076227A1 (zh) 人脸图像分类方法、装置及服务器
Jain et al. Visual assistance for blind using image processing
CN104143086A (zh) 人像比对在移动终端操作系统上的应用技术
WO2022089124A1 (zh) 证照真伪识别方法、装置、计算机可读介质及电子设备
WO2022021029A1 (zh) 检测模型训练方法、装置、检测模型使用方法及存储介质
TW200905577A (en) Iris recognition system
US20230082906A1 (en) Liveness detection method
WO2015131571A1 (zh) 一种实现图片排序的方法及终端
WO2018145246A1 (zh) 一种导盲系统
CN105095841A (zh) 生成眼镜的方法及装置
Satani et al. Ai powered glasses for visually impaired person
Mukherjee et al. Energy efficient face recognition in mobile-fog environment
Singh et al. IPCRF: An End-to-End Indian Paper Currency Recognition Framework for Blind and Visually Impaired People
Sarmah et al. Facial identification expression-based attendance monitoring and emotion detection—A deep CNN approach
CN114140839A (zh) 用于人脸识别的图像发送方法、装置、设备及存储介质

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: 17896086

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 1205 DATED 19/12/2019)

122 Ep: pct application non-entry in european phase

Ref document number: 17896086

Country of ref document: EP

Kind code of ref document: A1