WO2016155627A1 - 一种识别图片中的字符的方法和装置 - Google Patents

一种识别图片中的字符的方法和装置 Download PDF

Info

Publication number
WO2016155627A1
WO2016155627A1 PCT/CN2016/077865 CN2016077865W WO2016155627A1 WO 2016155627 A1 WO2016155627 A1 WO 2016155627A1 CN 2016077865 W CN2016077865 W CN 2016077865W WO 2016155627 A1 WO2016155627 A1 WO 2016155627A1
Authority
WO
WIPO (PCT)
Prior art keywords
character
picture
candidate
candidate character
probability
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2016/077865
Other languages
English (en)
French (fr)
Inventor
薛永刚
贾文杰
项碧波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Qihoo Technology Co Ltd
Qizhi Software Beijing Co Ltd
Original Assignee
Beijing Qihoo Technology Co Ltd
Qizhi Software Beijing Co Ltd
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 Beijing Qihoo Technology Co Ltd, Qizhi Software Beijing Co Ltd filed Critical Beijing Qihoo Technology Co Ltd
Publication of WO2016155627A1 publication Critical patent/WO2016155627A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/10—Image acquisition

Definitions

  • the present invention relates to the field of computer technologies, and in particular, to a method and apparatus for recognizing characters in a picture.
  • the present invention has been made in order to provide a method and apparatus for recognizing characters in a picture that overcomes the above problems or at least partially solves or alleviates the above problems.
  • a method of identifying a character in a picture comprising:
  • Identifying a character placeholder included in the picture obtaining a candidate character set corresponding to each character placeholder and a probability parameter corresponding to each candidate character;
  • one candidate character is selected from each candidate character set corresponding to each character placeholder to obtain a candidate character combination; multiple selections are performed, and different character combinations are selected each time. , obtaining multiple candidate character combinations;
  • the candidate characters with the highest probability are combined as the result of character recognition for the picture.
  • an apparatus for recognizing characters in a picture includes:
  • An obtaining unit configured to identify a character placeholder included in the picture, obtain a candidate character set corresponding to each character placeholder, and a probability parameter corresponding to each candidate character;
  • the pre-processing unit is adapted to select one candidate character from the candidate character set corresponding to each character placeholder in turn according to the order of the characters occupied by the picture to obtain a candidate character combination; Select different character combinations to get multiple candidate character combinations;
  • the identifying unit is adapted to calculate the probability of each of the obtained candidate character combinations, and combine the candidate characters with the highest probability as the character recognition result for the picture.
  • a computer program comprising computer readable code, when the computer readable code is run on a terminal device, causes the terminal device to perform the identification picture of any of the above The method of characters in .
  • a computer readable medium storing a computer program as described above is provided.
  • the probability of the candidate character set corresponding to the character placeholder included in the picture and the probability parameter corresponding to each candidate character is calculated, and the probability of all candidate character combinations that may be included in the picture is calculated, and the probability is the highest.
  • the candidate character combination is used as a technical solution for the character recognition result of the picture, and the image information is converted into text information that can be recognized and processed by the computer, which greatly improves the efficiency of the user in storing, retrieving and processing the character information in the picture.
  • FIG. 1 shows a flow chart of a method of identifying characters in a picture, in accordance with one embodiment of the present invention
  • FIG. 2 illustrates a method of identifying search keywords in accordance with one embodiment of the present invention.
  • FIG. 3 shows a schematic diagram of an apparatus for identifying characters in a picture, in accordance with one embodiment of the present invention
  • FIG. 4 shows a schematic diagram of an apparatus for identifying search keywords in accordance with one embodiment of the present invention
  • FIG. 5A shows a schematic diagram of a picture for character recognition according to an embodiment of the present invention
  • FIG. 5B shows a schematic diagram of a first picture according to another embodiment of the present invention.
  • FIG. 5C is a schematic diagram showing a second picture according to another embodiment of the present invention.
  • Figure 6 shows schematically a block diagram of a terminal device for carrying out the method according to the invention
  • Fig. 7 schematically shows a storage unit for holding or carrying program code implementing the method according to the invention.
  • FIG. 1 shows a flow chart of a method of identifying characters in a picture, in accordance with one embodiment of the present invention. As shown in Figure 1, the method includes:
  • Step S110 identifying a character placeholder included in the picture, acquiring a candidate character set corresponding to each character placeholder and a probability parameter corresponding to each candidate character.
  • Step S120 sequentially select one candidate character from the candidate character set corresponding to each character placeholder according to the order of each character occupying position of the picture to obtain a candidate character combination; perform multiple selections, and select different characters each time. Combine to get multiple candidate character combinations.
  • Step S130 calculating the probability of each of the obtained candidate character combinations.
  • step S140 the candidate characters with the highest probability are combined as the character recognition result for the picture.
  • the method shown in FIG. 1 obtains all candidate characters that may be included in the image by acquiring a candidate character set corresponding to the character placeholder included in the picture and a probability parameter corresponding to each candidate character. The combined probability is calculated, and the candidate character with the highest probability is combined as the character recognition result for the picture.
  • the program can convert the image information into text information that can be recognized and processed by the computer, which greatly improves the efficiency of the user in storing, retrieving and processing the character information in the picture.
  • step S110 of the method shown in FIG. 1 uses an optical character recognition (OCR) technology to identify a character occupying position included in a picture, and obtains a candidate character set corresponding to each character placeholder and The probability parameter corresponding to each candidate character.
  • OCR optical character recognition
  • the probability of each of the candidate character combinations obtained by step S130 of the method shown in FIG. 1 is: based on the n-gram model, the probability of each of the candidate character combinations obtained is calculated. Specifically, for each candidate character in a candidate character combination, a conditional probability that the candidate character appears under the condition that the first n-1 candidate characters have been determined is calculated according to the probability parameter of each candidate character; and the candidate character is calculated. The product of the conditional probabilities of the candidate characters in the combination is taken as the probability of the candidate character combination.
  • FIG. 5A is a schematic diagram of a picture for performing character recognition according to an embodiment of the present invention.
  • the character occupied by the picture is recognized, and each character place is acquired.
  • the corresponding candidate character set and the result of the probability parameter corresponding to each candidate character are as shown in Table 1:
  • Character placeholder a set of candidate characters corresponding to each character placeholder Probability parameter corresponding to each candidate character 1 Hand, season, care, year, abundance 35,49,51,53,57 2 Machine, ⁇ , ⁇ , ⁇ , maple 22,43,52,52,56 3 Tube, ⁇ , camp, camp, ⁇ 26,49,52,52,52 4 Home, ⁇ , elephant, ⁇ , frozen 23,61,81,82,83
  • one candidate character is selected from each candidate character set corresponding to each character placeholder to obtain a candidate character combination;
  • the probability of each candidate character combination is calculated based on the 4-ary grammar model, that is, for each candidate character in a candidate character combination, the probability parameter is calculated according to the probability parameter of each candidate character.
  • the probability of occurrence under the condition that its first three candidate characters have been determined is: P(hand
  • P (mobile phone housekeeper) P (hand
  • the calculation process of other candidate character combinations is the same, and will not be described again.
  • the candidate character with the highest probability is combined as the character recognition result for the picture.
  • the candidate character combination with the highest probability is “mobile phone housekeeper”, that is, the character recognition result of the picture shown in FIG. 5A.
  • FIG. 2 shows a flow chart of a method of identifying search keywords in accordance with one embodiment of the present invention. As shown in Figure 2, the method includes:
  • step S210 in response to the user's touch screen operation, the picture is intercepted according to the operation range to obtain the first picture; and the predetermined area is expanded according to the operation range to perform picture interception, and the second picture is obtained.
  • Step S220 respectively identifying characters in the first picture and the second picture to obtain corresponding character combinations.
  • Step S230 Select a character combination from the combination of characters corresponding to the first picture and the second picture as the search keyword after the recognition according to the preset policy.
  • the method shown in FIG. 2 intercepts two images with different range sizes in response to the user's touch screen operation, and selects two images from each of the two images by separately identifying and then processing the two images according to the preset strategy.
  • the recognition result of the letter is searched as a search keyword.
  • the solution has the following beneficial effects: Firstly, the recognition of the search keyword for the user's touch screen operation is realized, and the user does not need to manually input the search keyword, thereby simplifying the search operation process and meeting the user's needs; secondly, adopting Two ways to comprehensively identify related images, avoiding the missing information in a single picture or More than enough, the accuracy of character recognition in the picture is further improved, thereby improving the accuracy of identifying the search keyword.
  • step S220 of the method shown in FIG. 2 respectively identifies characters in the first picture and the second picture, and obtains corresponding character combinations as: by identifying the picture as described in any of the above embodiments.
  • the character method respectively identifies the characters in the first picture and the second picture to obtain a corresponding character combination.
  • step S220 of the method shown in FIG. 2 respectively identifies characters in the first picture and the second picture, and obtaining corresponding character combinations further includes: obtaining pixel coordinates of each character occupying position in the character combination. .
  • step S230 of the method shown in FIG. 2, according to a preset policy, selecting a character combination from the combination of characters corresponding to the first picture and the second picture as the search keyword includes:
  • step S231 in the character combination corresponding to the second picture, the character combination positions corresponding to the first picture are the same and the same length combination is used.
  • the second picture and the first picture are retained.
  • the corresponding character combination position is the same and the length is the same character combination.
  • Step S232 determining whether the average language model score of the character combination retained in the second picture is smaller than the average model score of the character combination corresponding to the first picture.
  • the average language model score of the character combination refers to the logarithm of the probability of the character combination, and the value obtained by averaging the number of characters in the character combination.
  • Step S233 yes, select a character combination corresponding to the first picture as a search keyword to perform a search; otherwise, select a character combination corresponding to the second picture as a search keyword to perform a search.
  • FIG. 5B is a schematic diagram showing a first picture according to another embodiment of the present invention
  • FIG. 5C is a schematic diagram showing a second picture according to another embodiment of the present invention, and the specific implementation shown in FIG. 5B and FIG. 5C
  • the image in response to the touch screen operation of the user, the image is intercepted according to the operation range, and the first picture as shown in FIG. 5B is obtained; and the predetermined area is expanded according to the operation range, and the picture is intercepted, and the second picture as shown in FIG. 5C is obtained.
  • Identifying the character placeholders included in the first picture obtaining a candidate character set corresponding to each character placeholder, a probability parameter corresponding to each candidate character, and a pixel coordinate of each character placeholder, and the result is shown in Table 2:
  • the probability of each candidate character combination is calculated based on the 4-ary grammar model, and the character combination with the highest probability of identifying the first picture is “Mobile Phone Easy”.
  • the specific recognition process has been described in detail in the foregoing, and will not be described again.
  • the character occupying position included in the second picture is identified, the candidate character set corresponding to each character placeholder, the probability parameter corresponding to each candidate character, and the pixel coordinate of each character placeholder are obtained, and the result is shown in Table 3. :
  • each candidate character combination is calculated based on the 4-ary grammar model, and the character combination with the highest probability of identifying the second picture is “t’ae. mobile phone housekeeper”.
  • the specific identification process has been described in detail in the foregoing, and will not be described again.
  • the "t'ae.” part of the character combination is located at a higher position in the second picture, and the "phone housekeeper” part is located at a lower position in the second picture.
  • the character combination corresponding to the first picture, "mobile phone tube” has the same character combination and the same length, according to "mobile phone management” and The pixel coordinate boundary of "t'ae. mobile phone housekeeper” and the pixel coordinates of each character placeholder, we can see that in the character combination "t'ae. mobile phone housekeeper” corresponding to the second picture, the “mobile phone housekeeper” part is The phone is easy to use in the same character group with the same length, so keep the "phone butler” character combination in the second picture.
  • the apparatus 300 for recognizing characters in a picture includes:
  • the obtaining unit 310 is adapted to identify a character placeholder included in the picture, obtain a candidate character set corresponding to each character placeholder, and a probability parameter corresponding to each candidate character.
  • the pre-processing unit 320 is adapted to select one candidate character from the candidate character set corresponding to each character placeholder in sequence according to the order of the characters occupied by the picture to obtain a candidate character combination; Select different combinations of characters to get multiple candidate character combinations.
  • the identifying unit 330 is adapted to calculate the obtained probability of each candidate character combination, and combine the candidate characters with the highest probability as the character recognition result for the picture.
  • the device shown in FIG. 3 acquires the candidate character set corresponding to the character placeholder included in the picture and the probability parameter corresponding to each candidate character through the mutual cooperation of the units, and the probability of all candidate character combinations that may be included in the picture.
  • the calculation is performed to combine the candidate characters with the highest probability as the character recognition result for the picture.
  • the program can convert image information into text information that can be recognized and processed by a computer, which greatly improves the efficiency of data storage, retrieval and processing by users.
  • the acquiring unit 310 of the apparatus shown in FIG. 3 is adapted to identify a character placeholder included in a picture by using an optical character recognition technology, obtain a candidate character set corresponding to each character placeholder, and each candidate The probability parameter corresponding to the character.
  • the identification unit 330 of the apparatus shown in FIG. 3 is adapted to be based on n A meta-grammar model that calculates the probability of each candidate combination of characters obtained.
  • the identifying unit 330 is adapted to calculate, for each candidate character in a candidate character combination, a condition that the candidate character appears under the condition that the first n-1 candidate characters have been determined according to the probability parameter of each candidate character. Probability; the product of the conditional probability of each candidate character in the candidate character combination is calculated as the probability of the candidate character combination.
  • the apparatus 400 for identifying a search keyword includes:
  • the image obtaining unit 410 is adapted to perform a picture capture according to the operation range in response to the user's touch screen operation to obtain a first picture; and further expand the predetermined area according to the operation range to perform picture interception to obtain a second picture.
  • the identification processing unit 420 is adapted to respectively identify the characters in the first picture and the second picture to obtain a corresponding character combination.
  • the search processing unit 430 is adapted to select a combination of characters from the combination of characters corresponding to the first picture and the second picture as the search keyword after the recognition according to the preset policy.
  • the device shown in FIG. 4 intercepts two images with different range sizes in response to the user's touch screen operation through the mutual cooperation of the units, and separately processes the two images according to the preset strategy, and then In the picture, select a more reliable recognition result as a search keyword to search.
  • the solution has the following beneficial effects: Firstly, the recognition of the search keyword for the user's touch screen operation is realized, and the user does not need to manually input the search keyword, thereby simplifying the search operation process and meeting the user's needs; secondly, adopting The method of comprehensively identifying two related pictures avoids the lack or surplus of information in a single picture, further improves the accuracy of character recognition in the picture, and further improves the accuracy of identifying the search keyword.
  • the identification processing unit 420 of the apparatus shown in FIG. 4 is adapted to respectively identify the first picture and the second picture by means of the apparatus 300 for recognizing characters in the picture as described in any of the above embodiments.
  • the characters in the box get the corresponding character combination.
  • identification processing unit 420 of the apparatus shown in FIG. 4 is further adapted to obtain pixel coordinates of each character occupying in the character combination.
  • the search processing unit 430 of the apparatus shown in FIG. 4 is adapted to In the character combination corresponding to the second picture, the character combination position corresponding to the first picture is the same and the length is the same character combination; determining whether the average language model score of the reserved character combination in the second picture is smaller than the character combination corresponding to the first picture The average model score; yes, the character combination corresponding to the first picture is selected as the search keyword for searching; otherwise, the character combination corresponding to the second picture is selected as the search keyword for searching.
  • the search processing unit 430 is adapted to retain, according to the pixel coordinate boundary of the character combination corresponding to the first picture and the second picture, and the pixel coordinates of each character placeholder, retain the second picture corresponding to the first picture.
  • the average language model score for the combination of characters refers to the logarithm of the probability of the combination of characters, the value obtained by averaging the number of characters in the combination of characters.
  • the technical solution provided by the present invention as a whole, on the one hand, obtains a candidate character set corresponding to a character placeholder included in a picture and a probability parameter corresponding to each candidate character, which may be included in the picture.
  • the probability of all candidate character combinations is calculated, and the candidate character with the highest probability is combined as the character recognition result for the picture.
  • two pictures with different range sizes are intercepted, and a more reliable recognition result is selected from the two pictures by separately identifying and then processing the two pictures according to the preset strategy. Search as a search keyword.
  • the scheme has the following beneficial effects: 1.
  • modules in the devices of the embodiments can be adaptively changed and placed in one or more devices different from the embodiment.
  • the modules or units or components of the embodiments may be combined into one module or unit or component, and further they may be divided into a plurality of sub-modules or sub-units or sub-components.
  • any combination of the features disclosed in the specification, including the accompanying claims, the abstract and the drawings, and any methods so disclosed, or All processes or units of the device are combined.
  • Each feature disclosed in this specification (including the accompanying claims, the abstract and the drawings) may be replaced by alternative features that provide the same, equivalent or similar purpose.
  • the various component embodiments of the present invention may be implemented in hardware, or in a software module running on one or more processors, or in a combination thereof.
  • a microprocessor or digital signal processor may be used in practice to implement some of some or all of the means for identifying characters in a picture or in accordance with an embodiment of the present invention. All features.
  • the invention can also be implemented as a device or device program (e.g., a computer program and a computer program product) for performing some or all of the methods described herein.
  • Such a program implementing the invention may be stored on a computer readable medium or may be in the form of one or more signals. Such signals may be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.
  • Figure 6 shows a terminal device in which the method according to the invention can be implemented.
  • the and terminal devices conventionally include a processor 610 and a computer program product or computer readable medium in the form of a memory 620.
  • the memory 620 may be an electronic memory such as a flash memory, an EEPROM (Electrically Erasable Programmable Read Only Memory), an EPROM, a hard disk, or a ROM.
  • Memory 620 has a memory space 630 for program code 631 for performing any of the method steps described above.
  • storage space 630 for program code may include various program code 631 for implementing various steps in the above methods, respectively.
  • the program code can be read from or written to one or more computer program products.
  • Such computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks.
  • Such a computer program product is typically a portable or fixed storage unit as described with reference to FIG.
  • the storage unit may have a storage section, a storage space, and the like arranged similarly to the storage 620 in the terminal device of FIG.
  • the program code can be compressed, for example, in an appropriate form.
  • the storage unit comprises computer readable code 631', ie code that can be read by a processor, such as 610, which when executed by the terminal device causes the terminal device to perform each of the methods described above step.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Character Discrimination (AREA)

Abstract

一种识别图片中的字符的方法和装置。所述方法包括：识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数(S110)；按照图片所包含的各字符占位的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，得到多个候选字符组合(S120)；计算所得到的每个候选字符组合的概率(S130)，将概率最高的候选字符组合作为对所述图片的字符识别结果(S140)。所述技术方案将图像信息完整、准确地转化为可以供计算机识别和处理的文本信息，自动提取图片中的字符信息，无需用户手动输入，满足用户需求。

Description

一种识别图片中的字符的方法和装置 技术领域
本发明涉及计算机技术领域，具体涉及一种识别图片中的字符的方法和装置。
背景技术
在信息技术及终端技术日益普及的今天，如何将字符方便、快捷地输入到终端中已经成为影响人机接口效率的一个重要难题。现有技术中，大部分用户仍然依赖于传统的键盘输入或手写输入来完成字符的输入，能够满足用户基本输入需求。但是，传统的输入方式也为用户带来了很多不便，例如，当用户对一个图片中所包含的字符有疑问，需要进行搜索时，需要将图片中所包含的字符依次手动输入到搜索栏中；或者，当用户需要保存一个图片中的电话号码时，也需要将图片中所包含的电话号码记在其他地方，再依次手动输入到电话薄中。可见，由于终端无法识别图片中所包含的字符，导致用户对图片中所包含的字符的处理十分费时费力，不符合用户的需求。
发明内容
鉴于上述问题，提出了本发明以便提供一种克服上述问题或者至少部分地解决或者减缓上述问题的一种识别图片中的字符的方法和装置。
依据本发明的一个方面，提供了一种识别图片中的字符的方法，该方法包括：
识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数；
按照所述图片所包含的各字符占位的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，得到多个候选字符组合；
计算所得到的每个候选字符组合的概率，
将概率最高的候选字符组合作为对所述图片的字符识别结果。
依据本发明的又一个方面，提供了一种识别图片中的字符的装置，该装 置包括：
获取单元，适于识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数；
预处理单元，适于按照所述图片所包含的各字符占位的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，得到多个候选字符组合；
识别单元，适于计算所得到的每个候选字符组合的概率，将概率最高的候选字符组合作为对所述图片的字符识别结果。
根据本发明的又一个方面，提供了一种计算机程序，包括计算机可读代码，当所述计算机可读代码在终端设备上运行时，导致所述终端设备执行如上任一项所述的识别图片中的字符的方法。
根据本发明的再一个方面，提供了一种计算机可读介质，其中存储了如上所述的计算机程序。
由上述可知，本发明这种通过获取图片所包含的字符占位对应的候选字符集合以及每个候选字符对应的概率参数，对图片中可能包含的所有候选字符组合的概率进行计算，将概率最高的候选字符组合作为对图片的字符识别结果的技术方案，将图像信息转化为可以供计算机识别和处理的文本信息，大大提高了用户对图片中的字符信息进行资料存储、检索、加工的效率。
上述说明仅是本发明技术方案的概述，为了能够更清楚了解本发明的技术手段，而可依照说明书的内容予以实施，并且为了让本发明的上述和其它目的、特征和优点能够更明显易懂，以下特举本发明的具体实施方式。
附图说明
通过阅读下文优选实施方式的详细描述，各种其他的优点和益处对于本领域普通技术人员将变得清楚明了。附图仅用于示出优选实施方式的目的，而并不认为是对本发明的限制。而且在整个附图中，用相同的参考符号表示相同的部件。在附图中：
图1示出了根据本发明一个实施例的一种识别图片中的字符的方法的流程图；
图2示出了根据本发明一个实施例的一种识别搜索关键词的方法 的流程图；
图3示出了根据本发明一个实施例的一种识别图片中的字符的装置的示意图；
图4示出了根据本发明一个实施例的一种识别搜索关键词的装置的示意图；
图5A示出了根据本发明一个实施例的用于进行字符识别的图片的示意图；
图5B示出了根据本发明另一个实施例的第一图片的示意图；
图5C示出了根据本发明另一个实施例的第二图片的示意图；
图6示意性地示出了用于执行根据本发明的方法的终端设备的框图；以及
图7示意性地示出了用于保持或者携带实现根据本发明的方法的程序代码的存储单元。
具体实施例
下面将参照附图更详细地描述本公开的示例性实施例。虽然附图中显示了本公开的示例性实施例，然而应当理解，可以以各种形式实现本公开而不应被这里阐述的实施例所限制。相反，提供这些实施例是为了能够更透彻地理解本公开，并且能够将本公开的范围完整的传达给本领域的技术人员。
图1示出了根据本发明一个实施例的一种识别图片中的字符的方法的流程图。如图1所示，该方法包括：
步骤S110，识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数。
步骤S120，按照图片所包含的各字符占位的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，得到多个候选字符组合。
步骤S130，计算所得到的每个候选字符组合的概率。
步骤S140，将概率最高的候选字符组合作为对图片的字符识别结果。
可见，图1所示的方法通过获取图片所包含的字符占位对应的候选字符集合以及每个候选字符对应的概率参数，对图片中可能包含的所有候选字符 组合的概率进行计算，将概率最高的候选字符组合作为对图片的字符识别结果。本方案可以将图像信息转化为可以供计算机识别和处理的文本信息，大大提高了用户对图片中的字符信息进行资料存储、检索、加工的效率。
在本发明的一个实施例中，图1所示方法的步骤S110利用光学字符识别(Optical Character Recognition，OCR)技术识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数。
在本发明的一个实施例中，图1所示方法的步骤S130计算所得到的每个候选字符组合的概率为：基于n元语法模型，计算所得到的每个候选字符组合的概率。具体为，对于一个候选字符组合中的每个候选字符，根据各候选字符的概率参数，计算该候选字符在其前n-1个候选字符已经确定的条件下出现的条件概率；计算该候选字符组合中各候选字符的条件概率的乘积，作为该候选字符组合的概率。
图5A示出了根据本发明一个实施例的用于进行字符识别的图片的示意图，在图5A所示的具体实施例中，首先，识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数的结果如表1所示：
表1
字符占位 每个字符占位对应的候选字符集合 每个候选字符对应的概率参数
1 手，季，乎，年，丰 35，49，51，53，57
2 机，杌，枧，视，枫 22，43，52，52，56
3 管，菅，营，營，眚 26，49，52，52，52
4 家，冢，象，彖，冻 23，61，81，82，83
如表1所示，图片中共有4个字符占位，每个字符占位对应的候选字符集合中包括5个候选字符。
接着，按照图5A所示图片中字符占位从左到右的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，能够得到54＝625个候选字符组合。
本实施例中，基于4元语法模型计算每个候选字符组合的概率，即对于一个候选字符组合中的每个候选字符，根据各候选字符的概率参数，计算该 候选字符在其前3个候选字符已经确定的条件下出现的条件概率；计算该候选字符组合中各候选字符的条件概率的乘积，作为该候选字符组合的概率。
具体地，以一个候选字符组合“手机管家”为例，为了表示图片中字符组合的完整性，要考虑字符组合的开始和结束，先将该字符组合处理为“O手机管家$”，“O”表示字符组合开始，“$”表示字符组合结束。对于“手机管家”中的“手”字，在其前3个候选字符已经确定的条件下出现的概率是：P(手|OOO)；对于“机”字，在其前3个候选字符已经确定的条件下出现的概率是：P(机|OO手)；对于“管”字，在其前3个候选字符已经确定的条件下出现的概率是：P(管|O手机)；对于“家”字，在其前3个候选字符已经确定的条件下出现的概率是：P(家|手机管)；对于“$”，在其前3个候选字符已经确定的条件下出现的概率是：P($|机管家)。
因此，“手机管家”的概率为：P(手机管家)＝P(手|OOO)×P(机|OO手)×P(管|O手机)×P(家|手机管)×P($|机管家)。其他候选字符组合的计算过程同理，不再赘述。将概率最高的候选字符组合作为对图片的字符识别结果，在本实施例中，概率最高的候选字符组合为“手机管家”，即为图5A所示图片的字符识别结果。
图2示出了根据本发明一个实施例的一种识别搜索关键词的方法的流程图。如图2所示，该方法包括：
步骤S210，响应于用户触屏操作，按照操作范围进行图片截取，获得第一图片；再按照操作范围扩大预定面积进行图片截取，获得第二图片。
步骤S220，分别识别第一图片和第二图片中的字符，得到相应的字符组合。
步骤S230，根据预设策略，从第一图片和第二图片对应的字符组合中选择一个字符组合作为识别后的搜索关键词。
可见，图2所示的方法响应于用户触屏操作截取了两张范围大小不同的图片，通过先分别识别再根据预设策略综合处理两张图片的方式，从两张图片中选择出更可信的识别结果作为搜索关键词进行搜索。与现有技术相比，本方案具有以下有益效果：首先，实现了对用户触屏操作的搜索关键词的识别，无需用户手动输入搜索关键词，简化搜索操作流程，符合用户需求；其次，采用了两张相关图片综合识别的方式，避免了单张图片中信息的缺失或 赘余，进一步提高了对图片中字符识别的准确率，进而提高了识别搜索关键词的准确率。
在本发明的一个实施例中，图2所示方法的步骤S220分别识别第一图片和第二图片中的字符，得到相应的字符组合为：通过如上文中任一实施例所述的识别图片中的字符的方法，分别识别第一图片和第二图片中的字符，得到相应的字符组合。
在本发明的另一个实施例中，图2所示方法的步骤S220分别识别第一图片和第二图片中的字符，得到相应的字符组合进一步包括：得到字符组合中各字符占位的像素坐标。
在本发明的一个实施例中，图2所示方法的步骤S230根据预设策略，从第一图片和第二图片对应的字符组合中选择一个字符组合作为搜索关键词包括：
步骤S231，在第二图片对应的字符组合中，保留与第一图片对应的字符组合位置相同且长度相同字符组合。
本步骤中，依据一个具体的实施例，具体为：根据第一图片和第二图片分别对应的字符组合的像素坐标边界和每个字符占位的像素坐标，保留第二图片中与第一图片对应的字符组合位置相同且长度相同字符组合。
步骤S232，判断第二图片中保留的字符组合的平均语言模型分数是否小于第一图片对应的字符组合的平均模型分数。
本步骤中，字符组合的平均语言模型分数是指：字符组合的概率的对数值，按照字符组合中的字符数取平均后得到的值。
步骤S233，是则，选择第一图片对应的字符组合作为搜索关键词进行搜索；否则，选择第二图片对应的字符组合作为搜索关键词进行搜索。
图5B示出了根据本发明另一个实施例的第一图片的示意图；图5C示出了根据本发明另一个实施例的第二图片的示意图，在图5B和图5C所示的具体的实施例中，响应于用户的触屏操作，按照操作范围进行图片截取，获得如图5B所示的第一图片；再按照操作范围扩大预定面积进行图片截取，获得如图5C所示的第二图片。识别第一图片所包含的字符占位，获取每个字符占位对应的候选字符集合、每个候选字符对应的概率参数和每个字符占位的像素坐标，结果如表2所示：
表2
Figure PCTCN2016077865-appb-000001
基于4元语法模型计算每个候选字符组合的概率，识别第一图片的概率最高的字符组合为“手机管易”，具体的识别过程在前文中已详细描述，不再赘述。
同理，识别第二图片所包含的字符占位，获取每个字符占位对应的候选字符集合、每个候选字符对应的概率参数和每个字符占位的像素坐标，结果如表3所示：
表3
Figure PCTCN2016077865-appb-000002
基于4元语法模型计算每个候选字符组合的概率，识别第二图片的概率最高的字符组合为“t’ae.手机管家”，具体的识别过程在前文中已详细描述，不再赘述。根据各字符占位对应的像素坐标可知，该字符组合中“t’ae.”部分位于第二图片中较高的位置，“手机管家”部分位于第二图片中较低的位置。
接着，在第二图片对应的字符组合“t’ae.手机管家”中，保留与第一图片对应的字符组合“手机管易”位置相同且长度相同的字符组合，根据“手机管易”和“t’ae.手机管家”的像素坐标边界和每个字符占位的像素坐标，可知，在第二图片对应的字符组合“t’ae.手机管家”中，“手机管家”部分是与“手机管易”位置相同且长度相同的字符组合，因此，保留第二图片中“手机管家”字符组合。
分别计算“手机管家”与“手机管易”的概率的对数值，按照字符组合中的字符数取平均后得到的值，得到ln[P(手机管家)]/4＞ln[P(手机管家)]/4，因此，选择第二图片对应的字符组合“手机管家”作为识别出的搜索关键词进行搜索。
图3示出了根据本发明一个实施例的一种识别图片中的字符的装置的示意图。如图3所示，该识别图片中的字符的装置300包括：
获取单元310，适于识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数。
预处理单元320，适于按照所述图片所包含的各字符占位的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，得到多个候选字符组合。
识别单元330，适于计算所得到的每个候选字符组合的概率，将概率最高的候选字符组合作为对所述图片的字符识别结果。
可见，图3所示的装置通过各单元的相互配合，获取图片所包含的字符占位对应的候选字符集合以及每个候选字符对应的概率参数，对图片中可能包含的所有候选字符组合的概率进行计算，将概率最高的候选字符组合作为对图片的字符识别结果。本方案可以将图像信息转化为可以供计算机识别和处理的文本信息，大大提高了用户进行资料存储、检索、加工的效率。
在本发明的一个实施例中，图3所示装置的获取单元310，适于利用光学字符识别技术识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数。
在本发明的一个实施例中，图3所示装置的识别单元330，适于基于n 元语法模型，计算所得到的每个候选字符组合的概率。具体为，识别单元330，适于对于一个候选字符组合中的每个候选字符，根据各候选字符的概率参数，计算该候选字符在其前n-1个候选字符已经确定的条件下出现的条件概率；计算该候选字符组合中各候选字符的条件概率的乘积，作为该候选字符组合的概率。
具体的实施例如图5A所在的实施例，上文中已详细说明，在此不再赘述。
图4示出了根据本发明一个实施例的一种识别搜索关键词的装置的示意图。如图4所示，该识别搜索关键词的装置400包括：
图片获取单元410，适于响应于用户触屏操作，按照所述操作范围进行图片截取，获得第一图片；再按照所述操作范围扩大预定面积进行图片截取，获得第二图片。
识别处理单元420，适于分别识别第一图片和第二图片中的字符，得到相应的字符组合。
搜索处理单元430，适于根据预设策略，从第一图片和第二图片对应的字符组合中选择一个字符组合作为识别后的搜索关键词。
可见，图4所示的装置通过各单元的相互配合，响应于用户触屏操作截取了两张范围大小不同的图片，通过先分别识别再根据预设策略综合处理两张图片的方式，从两张图片中选择出更可信的识别结果作为搜索关键词进行搜索。与现有技术相比，本方案具有以下有益效果：首先，实现了对用户触屏操作的搜索关键词的识别，无需用户手动输入搜索关键词，简化搜索操作流程，符合用户需求；其次，采用了两张相关图片综合识别的方式，避免了单张图片中信息的缺失或赘余，进一步提高了对图片中字符识别的准确率，进而提高了识别搜索关键词的准确率。
在本发明的一个实施例中，图4所示装置的识别处理单元420，适于通过如上文中任一实施例所述的识别图片中的字符的装置300，分别识别第一图片和第二图片中的字符，得到相应的字符组合。
进一步地，图4所示装置的识别处理单元420，还适于得到字符组合中各字符占位的像素坐标。
在本发明的一个实施例中，图4所示装置的搜索处理单元430，适于 在第二图片对应的字符组合中，保留与第一图片对应的字符组合位置相同且长度相同字符组合；判断第二图片中保留的字符组合的平均语言模型分数是否小于第一图片对应的字符组合的平均模型分数；是则，选择第一图片对应的字符组合作为搜索关键词进行搜索；否则，选择第二图片对应的字符组合作为搜索关键词进行搜索。其中，具体地，搜索处理单元430，适于根据第一图片和第二图片分别对应的字符组合的像素坐标边界和每个字符占位的像素坐标，保留第二图片中与第一图片对应的字符组合位置相同且长度相同字符组合。此外，在一个实施例中，字符组合的平均语言模型分数是指：字符组合的概率的对数值，按照字符组合中的字符数取平均后得到的值。
具体的实施例如图5B和图5C所在的实施例，上文中已详细说明，在此不再赘述。
综上所述，本发明提供的技术方案作为一个整体来看，一方面，通过获取图片所包含的字符占位对应的候选字符集合以及每个候选字符对应的概率参数，对图片中可能包含的所有候选字符组合的概率进行计算，将概率最高的候选字符组合作为对图片的字符识别结果。另一方面，响应于用户触屏操作截取了两张范围大小不同的图片，通过先分别识别再根据预设策略综合处理两张图片的方式，从两张图片中选择出更可信的识别结果作为搜索关键词进行搜索。与现有技术相比，本方案具有以下有益效果：1、利用自然语言n元语法模型修正光学字符识别技术本身的识别错误问题，优化效果；2、动态规划找到最优的候选字符组合，提升识别效果；3、采用大图和小图的综合比较算法，相互验证和补充，避免了单张图片中信息的缺失或赘余；4、基于小图像素坐标来选择最终识别结果的位置和长度。本发明提高了对图片中字符识别的准确率，进而提高了识别搜索关键词的准确率，无需用户手动输入，提高搜索效率，符合用户需求。
需要说明的是：
在此提供的算法和显示不与任何特定计算机、虚拟装置或者其它设备固有相关。各种通用装置也可以与基于在此的示教一起使用。根 据上面的描述，构造这类装置所要求的结构是显而易见的。此外，本发明也不针对任何特定编程语言。应当明白，可以利用各种编程语言实现在此描述的本发明的内容，并且上面对特定语言所做的描述是为了披露本发明的最佳实施方式。
在此处所提供的说明书中，说明了大量具体细节。然而，能够理解，本发明的实施例可以在没有这些具体细节的情况下实践。在一些实例中，并未详细示出公知的方法、结构和技术，以便不模糊对本说明书的理解。
类似地，应当理解，为了精简本公开并帮助理解各个发明方面中的一个或多个，在上面对本发明的示例性实施例的描述中，本发明的各个特征有时被一起分组到单个实施例、图、或者对其的描述中。然而，并不应将该公开的方法解释成反映如下意图：即所要求保护的本发明要求比在每个权利要求中所明确记载的特征更多的特征。更确切地说，如下面的权利要求书所反映的那样，发明方面在于少于前面公开的单个实施例的所有特征。因此，遵循具体实施方式的权利要求书由此明确地并入该具体实施方式，其中每个权利要求本身都作为本发明的单独实施例。
本领域那些技术人员可以理解，可以对实施例中的设备中的模块进行自适应性地改变并且把它们设置在与该实施例不同的一个或多个设备中。可以把实施例中的模块或单元或组件组合成一个模块或单元或组件，以及此外可以把它们分成多个子模块或子单元或子组件。除了这样的特征和/或过程或者单元中的至少一些是相互排斥之外，可以采用任何组合对本说明书(包括伴随的权利要求、摘要和附图)中公开的所有特征以及如此公开的任何方法或者设备的所有过程或单元进行组合。除非另外明确陈述，本说明书(包括伴随的权利要求、摘要和附图)中公开的每个特征可以由提供相同、等同或相似目的的替代特征来代替。
此外，本领域的技术人员能够理解，尽管在此所述的一些实施例包括其它实施例中所包括的某些特征而不是其它特征，但是不同实施例的特征的组合意味着处于本发明的范围之内并且形成不同的实施 例。例如，在下面的权利要求书中，所要求保护的实施例的任意之一都可以以任意的组合方式来使用。
本发明的各个部件实施例可以以硬件实现，或者以在一个或者多个处理器上运行的软件模块实现，或者以它们的组合实现。本领域的技术人员应当理解，可以在实践中使用微处理器或者数字信号处理器(DSP)来实现根据本发明实施例的一种识别图片中的字符的装置中的一些或者全部部件的一些或者全部功能。本发明还可以实现为用于执行这里所描述的方法的一部分或者全部的设备或者装置程序(例如，计算机程序和计算机程序产品)。这样的实现本发明的程序可以存储在计算机可读介质上，或者可以具有一个或者多个信号的形式。这样的信号可以从因特网网站上下载得到，或者在载体信号上提供，或者以任何其他形式提供。
例如，图6示出了可以实现根据本发明的方法的终端设备。该和终端设备传统上包括处理器610和以存储器620形式的计算机程序产品或者计算机可读介质。存储器620可以是诸如闪存、EEPROM(电可擦除可编程只读存储器)、EPROM、硬盘或者ROM之类的电子存储器。存储器620具有用于执行上述方法中的任何方法步骤的程序代码631的存储空间630。例如，用于程序代码的存储空间630可以包括分别用于实现上面的方法中的各种步骤的各个程序代码631。这些程序代码可以从一个或者多个计算机程序产品中读出或者写入到这一个或者多个计算机程序产品中。这些计算机程序产品包括诸如硬盘，紧致盘(CD)、存储卡或者软盘之类的程序代码载体。这样的计算机程序产品通常为如参考图7所述的便携式或者固定存储单元。该存储单元可以具有与图6的终端设备中的存储器620类似布置的存储段、存储空间等。程序代码可以例如以适当形式进行压缩。通常，存储单元包括计算机可读代码631’，即可以由例如诸如610之类的处理器读取的代码，这些代码当由终端设备运行时，导致该终端设备执行上面所描述的方法中的各个步骤。
本文中所称的“一个实施例”、“实施例”或者“一个或者多个实施例”意味着，结合实施例描述的特定特征、结构或者特性包括在 本发明的至少一个实施例中。此外，请注意，这里“在一个实施例中”的词语例子不一定全指同一个实施例。
应该注意的是上述实施例对本发明进行说明而不是对本发明进行限制，并且本领域技术人员在不脱离所附权利要求的范围的情况下可设计出替换实施例。在权利要求中，不应将位于括号之间的任何参考符号构造成对权利要求的限制。单词“包含”不排除存在未列在权利要求中的元件或步骤。位于元件之前的单词“一”或“一个”不排除存在多个这样的元件。本发明可以借助于包括有若干不同元件的硬件以及借助于适当编程的计算机来实现。在列举了若干装置的单元权利要求中，这些装置中的若干个可以是通过同一个硬件项来具体体现。单词第一、第二、以及第三等的使用不表示任何顺序。可将这些单词解释为名称。
此外，还应当注意，本说明书中使用的语言主要是为了可读性和教导的目的而选择的，而不是为了解释或者限定本发明的主题而选择的。因此，在不偏离所附权利要求书的范围和精神的情况下，对于本技术领域的普通技术人员来说许多修改和变更都是显而易见的。对于本发明的范围，对本发明所做的公开是说明性的，而非限制性的，本发明的范围由所附权利要求书限定。

Claims (10)

  1. 一种识别图片中的字符的方法，其中，该方法包括：
    识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数；
    按照所述图片所包含的各字符占位的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，得到多个候选字符组合；
    计算所得到的每个候选字符组合的概率，
    将概率最高的候选字符组合作为对所述图片的字符识别结果。
  2. 如权利要求1所述的方法，其中，所述计算所得到的每个候选字符组合的概率为：
    基于n元语法模型，计算所得到的每个候选字符组合的概率。
  3. 如权利要求1或2所述的方法，其中，所述基于n元语法模型，计算所得到的每个候选字符组合的概率包括：
    对于一个候选字符组合中的每个候选字符，根据各候选字符的概率参数，计算该候选字符在其前n-1个候选字符已经确定的条件下出现的条件概率；
    计算该候选字符组合中各候选字符的条件概率的乘积，作为该候选字符组合的概率。
  4. 如权利要求1-3中任一项所述的方法，其中，
    利用光学字符识别技术识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数。
  5. 一种识别图片中的字符的装置，其中，该装置包括：
    获取单元，适于识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数；
    预处理单元，适于按照所述图片所包含的各字符占位的顺序，依次从每个字符占位对应的候选字符集合中选取一个候选字符得到一个候选字符组合；进行多次选取，每次选取到不同的字符组合，得到多个候选字符组合；
    识别单元，适于计算所得到的每个候选字符组合的概率，将概率最高 的候选字符组合作为对所述图片的字符识别结果。
  6. 如权利要求5所述的装置，其中，
    所述识别单元，适于基于n元语法模型，计算所得到的每个候选字符组合的概率。
  7. 如权利要求5或6所述的装置，其中，
    所述识别单元，适于对于一个候选字符组合中的每个候选字符，根据各候选字符的概率参数，计算该候选字符在其前n-1个候选字符已经确定的条件下出现的条件概率；再计算该候选字符组合中各候选字符的条件概率的乘积，作为该候选字符组合的概率。
  8. 如权利要求5-7中任一项所述的装置，其中，
    所述获取单元，适于利用光学字符识别技术识别图片所包含的字符占位，获取每个字符占位对应的候选字符集合以及每个候选字符对应的概率参数
  9. 一种计算机程序，包括计算机可读代码，当所述计算机可读代码在终端设备上运行时，导致所述终端设备执行根据权利要求1-4中的任一个所述的识别图片中的字符的方法。
  10. 一种计算机可读介质，其中存储了如权利要求9所述的计算机程序。
PCT/CN2016/077865 2015-04-03 2016-03-30 一种识别图片中的字符的方法和装置 Ceased WO2016155627A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510158353.9 2015-04-03
CN201510158353.9A CN104766077B (zh) 2015-04-03 2015-04-03 一种识别图片中的字符的方法和装置

Publications (1)

Publication Number Publication Date
WO2016155627A1 true WO2016155627A1 (zh) 2016-10-06

Family

ID=53647891

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/077865 Ceased WO2016155627A1 (zh) 2015-04-03 2016-03-30 一种识别图片中的字符的方法和装置

Country Status (2)

Country Link
CN (1) CN104766077B (zh)
WO (1) WO2016155627A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111444906A (zh) * 2020-03-24 2020-07-24 腾讯科技（深圳）有限公司 基于人工智能的图像识别方法和相关装置
CN111914825A (zh) * 2020-08-03 2020-11-10 腾讯科技（深圳）有限公司 文字识别方法、装置及电子设备
CN112232347A (zh) * 2020-09-07 2021-01-15 平安科技（深圳）有限公司 基于概率矩阵的字符识别方法、装置、设备及存储介质
CN113221890A (zh) * 2021-05-25 2021-08-06 深圳市瑞驰信息技术有限公司 一种基于ocr的云手机文字内容监管方法和系统和系统
CN114677681A (zh) * 2018-02-06 2022-06-28 创新先进技术有限公司 用于数据标注的方法、装置和电子设备

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104766077B (zh) * 2015-04-03 2017-04-12 北京奇虎科技有限公司 一种识别图片中的字符的方法和装置
CN105069898B (zh) * 2015-07-15 2018-03-30 广州敦和信息技术有限公司 一种发票字条自动定位的方法及装置
CN109344830B (zh) * 2018-08-17 2024-06-28 平安科技（深圳）有限公司 语句输出、模型训练方法、装置、计算机设备及存储介质
JP6759306B2 (ja) * 2018-11-26 2020-09-23 キヤノン株式会社 画像処理装置及びその制御方法、プログラム
CN111507344A (zh) * 2019-01-30 2020-08-07 北京奇虎科技有限公司 从图像中识别出文字的方法和装置
CN110414450A (zh) * 2019-07-31 2019-11-05 北京字节跳动网络技术有限公司 关键词检测方法、装置、存储介质及电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090208107A1 (en) * 2008-02-19 2009-08-20 Fujitsu Limited Character recognition apparatus and character recognition method
CN102024139A (zh) * 2009-09-18 2011-04-20 富士通株式会社 字符串识别装置和方法
CN102982330A (zh) * 2012-11-21 2013-03-20 新浪网技术(中国)有限公司 文字图像中字符识别方法和识别装置
CN104766077A (zh) * 2015-04-03 2015-07-08 北京奇虎科技有限公司 一种识别图片中的字符的方法和装置

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102298486A (zh) * 2010-06-22 2011-12-28 广东国笔科技股份有限公司 一种基于触摸屏的快速调用系统及方法
CN102750272B (zh) * 2012-07-02 2015-01-14 安徽科大讯飞信息科技股份有限公司 字符手写输入候选项优化方法及系统
CN102799379A (zh) * 2012-07-13 2012-11-28 北京小米科技有限责任公司 一种字符输入方法及字符输入设备
CN103076892B (zh) * 2012-12-31 2016-09-28 百度在线网络技术（北京）有限公司 一种用于提供输入字符串所对应的输入候选项的方法与设备
CN104360759B (zh) * 2014-11-21 2017-03-08 百度在线网络技术（北京）有限公司 候选字排序方法、装置和文字输入方法、设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090208107A1 (en) * 2008-02-19 2009-08-20 Fujitsu Limited Character recognition apparatus and character recognition method
CN102024139A (zh) * 2009-09-18 2011-04-20 富士通株式会社 字符串识别装置和方法
CN102982330A (zh) * 2012-11-21 2013-03-20 新浪网技术(中国)有限公司 文字图像中字符识别方法和识别装置
CN104766077A (zh) * 2015-04-03 2015-07-08 北京奇虎科技有限公司 一种识别图片中的字符的方法和装置

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114677681A (zh) * 2018-02-06 2022-06-28 创新先进技术有限公司 用于数据标注的方法、装置和电子设备
CN111444906A (zh) * 2020-03-24 2020-07-24 腾讯科技（深圳）有限公司 基于人工智能的图像识别方法和相关装置
CN111444906B (zh) * 2020-03-24 2023-09-29 腾讯科技（深圳）有限公司 基于人工智能的图像识别方法和相关装置
CN111914825A (zh) * 2020-08-03 2020-11-10 腾讯科技（深圳）有限公司 文字识别方法、装置及电子设备
CN111914825B (zh) * 2020-08-03 2023-10-27 腾讯科技（深圳）有限公司 文字识别方法、装置及电子设备
CN112232347A (zh) * 2020-09-07 2021-01-15 平安科技（深圳）有限公司 基于概率矩阵的字符识别方法、装置、设备及存储介质
CN112232347B (zh) * 2020-09-07 2023-06-20 平安科技（深圳）有限公司 基于概率矩阵的字符识别方法、装置、设备及存储介质
CN113221890A (zh) * 2021-05-25 2021-08-06 深圳市瑞驰信息技术有限公司 一种基于ocr的云手机文字内容监管方法和系统和系统

Also Published As

Publication number Publication date
CN104766077A (zh) 2015-07-08
CN104766077B (zh) 2017-04-12

Similar Documents

Publication Publication Date Title
CN104766077B (zh) 一种识别图片中的字符的方法和装置
US12038977B2 (en) Visual recognition using user tap locations
US8649614B2 (en) Method of compiling image database, image database compilation program, image retrieval method
CN110555372A (zh) 数据录入方法、装置、设备及存储介质
JP5615932B2 (ja) 検索方法およびシステム
WO2021174717A1 (zh) 文本意图识别方法、装置、计算机设备和存储介质
US10482146B2 (en) Systems and methods for automatic customization of content filtering
US20150169978A1 (en) Selection of representative images
WO2015184992A1 (zh) 一种识别重复图片的方法、图片搜索去重方法及其装置
WO2019080411A1 (zh) 电子装置、人脸图像聚类搜索方法和计算机可读存储介质
US20140289632A1 (en) Picture drawing support apparatus and method
CN106156794B (zh) 基于文字风格识别的文字识别方法及装置
WO2015197029A1 (zh) 一种人脸相似度识别方法和系统
WO2016015621A1 (zh) 人脸图片人名识别方法和系统
CN110163061B (zh) 用于提取视频指纹的方法、装置、设备和计算机可读介质
US20240348846A1 (en) Video generating method and apparatus, electronic device, and readable storage medium
CN108595679B (zh) 一种标签确定方法、装置、终端和存储介质
WO2016045378A1 (zh) 一种网页分类方法及装置
CN110381392A (zh) 一种视频摘要提取方法及其系统、装置、存储介质
CN104572717A (zh) 信息搜索方法和装置
CN106649629B (zh) 一种关联书本与电子资源的系统
CN106844381B (zh) 图像处理装置及方法
CN115269765A (zh) 账号识别方法、装置、电子设备和存储介质
WO2016058520A1 (zh) 人脸图片人名识别方法和装置
CN102760127B (zh) 基于扩展文本信息来确定资源类型的方法、装置及设备

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: 16771382

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16771382

Country of ref document: EP

Kind code of ref document: A1