A method, device and electronic device for processing image information

Abstract

The present invention discloses a method, device and electronic device for processing image information, the method comprising: performing repeated image clustering on images in a web page to obtain a set of tagged text domains corresponding to each cluster; for cluster images of each cluster, obtaining keywords and word weights of keywords contained in each tagged text domain in the tagged text domain set according to the tagged text domain set; screening the tagged text domains in the tagged text domain set according to the word weights of the keywords in the tagged text domains and the word frequencies of the keywords in the corresponding tagged text domains to obtain multiple candidate text domains; removing similar texts from the multiple candidate text domains, and obtaining description text information of the cluster images according to the multiple candidate text domains after removing similar texts. The above scheme obtains a set of tagged text domains by performing repeated image clustering, and selects high-quality text domains from them as image description text information, thereby solving the technical problem of low accuracy of image description text information in the prior art.

Description

Technical Field

[0001] The present invention relates to the field of software technology, and in particular to a method, device and electronic equipment for processing picture information. Background Art

[0002] In Internet applications, image searches require mining text description information for Internet images so that image search results can be provided based on the match between query terms and text description information. Therefore, the accuracy of text description information will directly affect the accuracy of image search results.

[0003] At present, a large number of pictures are added to the Internet every day. Usually, the added pictures will have some tagged text fields, such as the picture title, the surrounding text of the picture, etc., which form the text description information of the picture. However, as the pictures spread and forward, the corresponding tagged text fields are often edited many times, resulting in the tagged text fields corresponding to the pictures being gradually distorted as the pictures spread, and even a large number of pictures and texts are inconsistent. In this case, how to mine accurate text description information for Internet pictures has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] The embodiments of the present invention provide a method, device and electronic device for processing image information, which are used to solve the technical problem of reduced accuracy of image text description information caused by propagation distortion of image tag text domain in the prior art, and improve the accuracy of image text description information.

[0005] An embodiment of the present invention provides a method for processing image information, the method comprising:

[0006] Repeatedly clustering the images in the webpage to obtain cluster images of each cluster and a set of marked text fields of the cluster images;

[0007] For each cluster image, obtaining keywords and word weights of the keywords contained in each tagged text domain in the tagged text domain set according to the tagged text domain set;

[0008] According to the word weight of the keyword in the marked text domain and the word frequency of the keyword in the corresponding marked text domain, the marked text domain in the marked text domain set is screened to obtain a plurality of candidate text domains;

[0009] Similar texts are removed from the multiple candidate text domains, and descriptive text information of the cluster image is obtained according to the multiple candidate text domains after the similar texts are removed.

[0010] Optionally, the step of obtaining the keywords and word weights of the keywords contained in each tagged text field in the tagged text field set according to the tagged text field set includes:

[0011] Obtaining keywords in each of the tag text fields;

[0012] The following target parameters are obtained for each keyword: the word frequency and the number of occurrences of the keyword in the corresponding markup text domain set and the number of website domain names corresponding to the keyword;

[0013] According to the target parameter of each keyword, the word weight of each keyword is calculated.

[0014] Optionally, calculating and obtaining a word weight of each keyword according to the target parameter of each keyword includes:

[0015] For each of the keywords, according to the word frequency and the number of occurrences of the keyword in each of the markup text domains, the importance of the keyword in each of the markup text domains is calculated, and the importance is accumulated by decaying according to the word frequency and the number of occurrences;

[0016] The word weight of each keyword is calculated based on the importance of the keyword in all the corresponding markup text domains and the number of website domain names corresponding to the keyword.

[0017] Optionally, the step of screening the tagged text domains in the tagged text domain set according to the word weights of the keywords in the tagged text domains and the word frequencies of the keywords in the corresponding tagged text domains to obtain a plurality of candidate text domains includes:

[0018] Calculate and obtain the score of each marked text field according to the word weight of the keyword in the marked text field and the word frequency of the keyword in the corresponding marked text field;

[0019] Obtaining the first N marked text fields with the largest scores in the marked text field set as reference text fields;

[0020] The plurality of candidate text domains are obtained according to the N reference text domains.

[0021] Optionally, obtaining the plurality of candidate text domains according to the N reference text domains includes:

[0022] Obtaining the text field length of each of the reference text fields;

[0023] According to the text domain type to which each of the reference text domains belongs, obtaining a domain weight of each of the reference text domains;

[0024] The N reference text domains are screened according to the text domain length and the domain weight to obtain the multiple candidate text domains.

[0025] Optionally, before screening the mark text fields in the mark text field set, the method further includes:

[0026] Keyword truncation is performed on the tag text fields in the tag text field set, and keywords after the Mth position in the tag text field that are sorted from large to small in terms of word weight are removed.

[0027] The embodiment of the present invention further provides a device for processing image information, the device comprising:

[0028] A clustering unit, used for clustering repeated images in the webpage, obtaining cluster images of each cluster and a set of marked text fields of the cluster images;

[0029] A weight acquisition unit, used for acquiring, for each cluster image, keywords contained in each tagged text domain in the tagged text domain set and word weights of the keywords according to the tagged text domain set;

[0030] A screening unit, configured to screen the marked text domains in the marked text domain set according to the word weights of the keywords in the marked text domains and the word frequencies of the keywords in the corresponding marked text domains, to obtain a plurality of candidate text domains;

[0031] The similarity removal unit is used to remove similar texts from the multiple candidate text domains, and obtain the description text information of the cluster image according to the multiple candidate text domains after the similar texts are removed.

[0032] Optionally, the weight acquisition unit is used to:

[0033] Obtaining keywords in each of the tag text fields;

[0034] The following target parameters are obtained for each keyword: the word frequency and the number of occurrences of the keyword in the corresponding markup text domain set and the number of website domain names corresponding to the keyword;

[0035] According to the target parameter of each keyword, the word weight of each keyword is calculated.

[0036] Optionally, the weight acquisition unit is further used for:

[0037] For each of the keywords, according to the word frequency and the number of occurrences of the keyword in each of the markup text domains, the importance of the keyword in each of the markup text domains is calculated, and the importance is accumulated by decaying according to the word frequency and the number of occurrences;

[0038] The word weight of each keyword is calculated based on the importance of the keyword in all the corresponding markup text domains and the number of website domain names corresponding to the keyword.

[0039] Optionally, the screening unit is used to:

[0040] Calculate and obtain the score of each marked text field according to the word weight of the keyword in the marked text field and the word frequency of the keyword in the corresponding marked text field;

[0041] Obtaining the first N marked text fields with the largest scores in the marked text field set as reference text fields;

[0042] The plurality of candidate text domains are obtained according to the N reference text domains.

[0043] Optionally, the screening unit is further used for:

[0044] Obtaining the text field length of each of the reference text fields;

[0045] According to the text domain type to which each of the reference text domains belongs, obtaining a domain weight of each of the reference text domains;

[0046] The N reference text domains are screened according to the text domain length and the domain weight to obtain the multiple candidate text domains.

[0047] Optionally, the device further comprises:

[0048] The keyword truncation unit is used to perform keyword truncation on the tagged text fields in the tagged text field set before screening the tagged text fields in the tagged text field set, and remove the keywords after the Mth position in the tagged text field whose word weights are ranked from large to small.

[0049] The above one or more technical solutions in the embodiments of the present application have at least the following technical effects:

[0050] The embodiment of the present application provides a method for processing image information, which performs repeated image clustering on images in a web page, obtains cluster images of each cluster and a set of tagged text domains of the cluster images; for each cluster image, obtains the keywords and the word weights of the keywords contained in each tagged text domain in the tagged text domain set according to the tagged text domain set; screens the tagged text domains in the tagged text domain set according to the word weights of the keywords in the tagged text domains and the word frequencies of the keywords in the corresponding tagged text domains to obtain multiple candidate text domains; removes similar texts from the multiple candidate text domains, and obtains the description text information of the cluster images according to the multiple candidate text domains after removing the similar texts. The tagged text domain set of the image is obtained by repeated image clustering, and the word weights of the image keywords are obtained accordingly, so that the tagged text domains are screened according to the word weights and text similarities to obtain more accurate image description text information, which solves the technical problem of reduced accuracy of image text description information caused by propagation distortion of image tagged text domains in the prior art, and improves the accuracy of image text description information. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of a process flow of a method for processing image information provided in an embodiment of the present application;

[0052] Figure 2 A block diagram of a picture information processing device provided in an embodiment of the present application;

[0053] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] In the technical solution provided in the embodiment of the present application, a method for processing image information is provided, which obtains a large number of tagged text domains by repeated image clustering, and selects high-quality text domains from them as text description information of the image, so as to solve the technical problem of reduced accuracy of new information in the image text description caused by distortion of propagation of image tagged text domains in the prior art.

[0055] The main implementation principles, specific implementation methods and corresponding beneficial effects of the technical solutions of the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0056] Example

[0057] Please refer to Figure 1 , the embodiment of the present application provides a method for processing image information, the method comprising the following S11~S14:

[0058] S11. Repeatedly cluster the images in the webpage to obtain cluster images of each cluster and a set of marked text fields of the cluster images.

[0059] Among them, duplicate image clustering is to group completely identical images together. For example, currently through image search, the total amount of original images searched before clustering is about 6 billion, and there are 800 million clusters after clustering. Each cluster represents a different image, that is, there are currently 800 million different images. The image corresponding to each cluster is called a cluster image. Each cluster image may appear in a large number of different web pages. The marked text domain of the cluster image can be obtained from each web page where it is located, such as the title of the image, the surrounding text of the image, the description text of the image, etc. Generally, the description text of the image is a line of text information adjacent to the image directly below the image, and the surrounding text of the image is text information within a certain distance from the image. All marked text domains constitute the marked text domain set of the cluster image.

[0060] S12: for each cluster image, obtain the keywords and word weights of the keywords contained in each tagged text field in the tagged text field set according to the tagged text field set.

[0061] The word weight of a keyword is used to reflect the distribution of the keyword in the tag text domain set. The larger the word weight, the more tag text domains in the tag text domain set contain the keyword, and the more important the keyword is to the cluster image. Conversely, the smaller the word weight, the fewer tag text domains in the tag text domain set contain the keyword, and the less important the keyword is to the cluster image. By obtaining the word weight of a keyword, the importance of the keyword to the cluster image can be clearly determined.

[0062] S13: Screen the tagged text domains in the tagged text domain set according to the word weights of the keywords in the tagged text domains and the word frequencies of the keywords in the corresponding tagged text domains to obtain a plurality of candidate text domains.

[0063] The candidate text domain is a high-quality text with a high keyword weight, a high keyword frequency, and a high domain weight in the tag text domain set. When screening the tag text domain, the tag text domain can be screened for the first time based on the word weight and word frequency, and then the tag text domain after the first screening can be screened for the second time based on the domain weight and text domain length of the text domain to reduce the amount of calculation for the second screening. The tag text domain length refers to the total number of keywords in the text domain.

[0064] S14, removing similar texts from the multiple candidate text domains, and obtaining descriptive text information of the cluster image according to the multiple candidate text domains after removing similar texts.

[0065] When selecting the text domain, the embodiment of the present application not only considers the relevance of the extracted text domain to the image, but also considers that the extracted text domain must have diversity, that is, low redundancy. If most of the extracted text domains are similar, then the diversity of information is lacking, so similar texts are removed for the multiple candidate text domains obtained by screening. The reference text domain after removing similar texts is used as the descriptive text information of the corresponding cluster image, or the keywords ranked in the top n positions by word weight are extracted from the reference text domain after removing similar texts as the descriptive text information of the corresponding cluster image.

[0066] In the specific real-time process, when S11 clusters duplicate images in web pages, it can first crawl images in web pages on the Internet; then, use the ring splitting algorithm to store each image (because hundreds of millions of images cannot exist on one machine, multiple machines will be used to store these images); then, perform a neighbor search on all images on all machines at the same time for each image to find images that are exactly the same as it, and then merge the neighbor search results of the image using the union-find set, so that all the identical images of each image are associated, and define an exclusive cluster id, thereby realizing the clustering of duplicate images, and the image corresponding to the cluster id is the cluster image of the cluster. After completing the clustering of duplicate images, S11 further crawls the labeled text domain of the image from the web pages to which a large number of duplicate images belong, and finally obtains the labeled text domain set of the cluster image.

[0067] After obtaining the set of tagged text domains for each cluster image based on repeated image clustering, S12 is executed to calculate the word weights of the keywords in the text domains, and the word weights are used to characterize the importance of the keywords to the cluster images. The importance of the keywords is proportional to their word weights. Specifically, for each set of tagged text domains for each cluster image, the keywords in each tagged text domain in the tagged text domain set are obtained; for each keyword, the following target parameters are obtained: the word frequency of the keyword in the tagged text domain set, the number of times the keyword appears in the tagged text domain to which it belongs, and the number of website domain names corresponding to the keyword; based on the target parameters of each keyword, the word weight of each keyword is calculated.

[0068] Among them, for each keyword, the importance of the keyword in each tag text domain is calculated based on the word frequency and number of occurrences of the keyword in each tag text domain to which it belongs. The importance is accumulated according to the frequency and number of occurrences of the keyword, which can be calculated by the following formula 2: Further, according to the importance of the keyword in all the tag text domains to which it belongs and the number of website domain names corresponding to the keyword, the word weight of each keyword is calculated. It can be calculated by the following formula 1:

[0069] Formula 1

[0070] Formula 2

[0071] w represents the word weight, i represents the i-th tagged text domain in the tagged text domain set, n represents the number of text domains in the tagged text domain set, f(i) represents the importance of the keyword in the i-th tagged text domain, n_page_domain uniq represents the number of website domain names corresponding to the cluster image, Max_weight represents the preset maximum weight value, norm_k represents the normalization parameter, m represents the word frequency, n domain Indicates the number of times the keyword appears in the tag text domain to which it belongs. The maximum value of the word weight can be 255, indicating that the keyword term frequently appears in each tag text domain, and the minimum value can be 1, indicating that it rarely appears in the cluster. The value of norm_k can be 4.

[0072] After obtaining the word weights in S12, before performing the tagged text domain screening in S13, the tagged text domain set may be preprocessed, including deduplication of the tagged text domain and / or keyword truncation. In this embodiment, considering that the keywords extracted by clustering have a long-tail effect, a large number of keywords with low weights will affect the effect of text domain screening. For this reason, the tagged text domains in the tagged text domain set are keyword truncated to remove the keywords after the Mth position in the tagged text domain in descending order of word weights. M is a preset value, which can be set according to experimental data.

[0073] Through keyword truncation, on the one hand, it ensures that the number of keywords in each marked text field does not exceed the preset value, and on the other hand, keywords with smaller word weights are removed to improve the efficiency and quality of text field screening.

[0074] For the preprocessed marked text domain set, execute S13 to perform text domain screening. Specifically, the text domain screening can be performed by sorting the marked text domain twice, first performing a rough sort, and then fine sorting the front part of the rough sort recall result. During the rough sorting, the score of each marked text domain in the marked text domain set is calculated based on the word weight of the keyword in the marked text domain and the word frequency of the keyword in the corresponding marked text domain; the first N marked text domains with the largest scores in the marked text domain set are obtained as reference text domains; during the fine sorting, multiple candidate text domains are obtained based on the obtained N reference text domains.

[0075] For rough sorting, the Okapi BM25 (BM25 for short) algorithm can be used to calculate the score of each labeled text field:

[0076] 1) For each keyword in the marked text field, count the frequency of each keyword appearing in the marked text field;

[0077] 2) weighting each keyword according to the word weight calculated in S12 and combining the word frequency of each keyword;

[0078] 3) According to 1) and 2), combined with the constants set in BM25, the score for marking the current text field is obtained.

[0079] According to the score of each marked text domain obtained by BM25 calculation, all marked text domains in the marked text domain set are sorted from large to small according to the score, and the first N marked text domains with the largest scores in the marked text domain set are obtained as reference text domains.

[0080] When fine sorting, the text domain length of each reference text domain can be obtained first; according to the text domain type to which each reference text domain belongs, the domain weight of each reference text domain is obtained; according to the text domain length and domain weight, N reference text domains are screened to obtain multiple candidate text domains. Among them, the text domain length can specifically include the number of characters contained in the text domain, and / or the number of keywords contained in the text domain. The domain weight of the text domain is a priori theoretical value. Since the selected text domain may come from different types of text domains such as title, anchor text anchor or body description desc, the quality of each type of text domain often varies. For example, the desc domain is generally better than the anchor domain. When selecting text description information, the desc domain will be selected. For this reason, there is an a priori theoretical value for different types of text domains to characterize the quality of this type of text.

[0081] In the specific real-time process, based on the rough sorting score, another score is made in combination with the domain weight and text domain length of each reference text domain, or in combination with the domain weight and text domain length of each reference text domain and the frequency of the keywords appearing in the corresponding text domain. For example, different features are weighted to obtain the score of each reference text domain, and fine sorting is performed based on the score.

[0082] After obtaining multiple candidate tagged text domains through screening in S13, S14 is further executed to remove similar texts. When removing similar texts, MMR (Maximal Marginal Relevance) can be used to reorder the texts of multiple candidate text domains. When MMR is reordering, the similarity is first measured. The similarity measurement can be Jaccard similarity and minimum common subsequence. If the similarity is too high, for example, greater than or equal to the similarity threshold, then one of the text domains is removed.

[0083] Furthermore, S14 can also perform a de-inclusion operation to reduce the redundancy of information. For example, assuming that the marked text domains obtained by BM25 include "Celebrity A and Celebrity B show affection" and "Celebrity A and Celebrity B show affection Sohu News", the latter text domain actually contains the content of the former text domain. Similar results need to be de-duplicated and containment relationship judgment, so that only the former marked text domain can be retained.

[0084] In the above embodiment, the picture description text information acquisition scheme based on repeated picture clustering can cluster pictures with exactly the same image content on pictures of the tens of billions of scale, and also aggregate all the texts of the cluster. Based on all the texts of the cluster, the high-quality text domain that best describes the cluster (the picture) is selected as its description text information, thereby improving the accuracy of picture search. For example, an original picture of star A and star B showing a heart on the Internet appears in different sites after multiple disseminations on the network. In the results crawled from different sources, the text domain of the picture has also changed. Not only have multiple stars been added, but the content has also weakened the importance of the keywords "star A" and "star B". Its marked text domain includes "Happy Valentine's Day", "star C and star D", "Valentine's Day stars take photos in fancy ways", etc. The high-quality text domain selected by the above method is "star A and star B show their love with difficulty" as the text description information of the cluster, rather than the text domain "star C and star D". After that, when the user searches for "star C and star D", the picture will not appear, thereby improving the accuracy of picture search.

[0085] Based on the same inventive concept, a method for processing image information is provided for the above embodiment. The present application embodiment also provides a device for processing image information. Please refer to Figure 2 , the device comprises:

[0086] A clustering unit 21 is used to perform repeated image clustering on the images in the web page, and obtain a cluster image of each cluster and a set of marked text fields of the cluster image;

[0087] A weight acquisition unit 22, configured to acquire, for each cluster image, keywords contained in each tagged text domain in the tagged text domain set and word weights of the keywords according to the tagged text domain set;

[0088] A screening unit 23, configured to screen the tagged text domains in the tagged text domain set according to the word weights of the keywords in the tagged text domains and the word frequencies of the keywords in the corresponding tagged text domains, to obtain a plurality of candidate text domains;

[0089] The similarity removal unit 24 is used to remove similar texts from the multiple candidate text fields, and obtain the description text information of the cluster image according to the multiple candidate text fields after the similar texts are removed.

[0090] As an optional implementation, the weight acquisition unit 22 is used to: obtain the keywords in each of the marked text domains; obtain the following target parameters for each keyword: the word frequency and number of occurrences of the keyword in the set of marked text domains to which it belongs and the number of website domain names corresponding to the keyword; and calculate the word weight of each of the keywords based on the target parameters of each of the keywords.

[0091] As an optional implementation, the weight acquisition unit 22 is also used to: for each of the keywords, calculate the importance of the keyword in each of the tag text domains to which it belongs according to the word frequency and the number of occurrences of the keyword in each of the tag text domains to which it belongs, and the importance is accumulated by decaying according to the word frequency and the number of occurrences; calculate the word weight of each of the keywords according to the importance of the keyword in all the tag text domains to which it belongs and the number of website domain names corresponding to the keyword.

[0092] As an optional implementation manner, the screening unit 23 is used to: calculate the score of each marked text domain according to the word weight of the keyword in the marked text domain and the word frequency of the keyword in the corresponding marked text domain; obtain the first N marked text domains with the largest scores in the marked text domain set as reference text domains; and obtain the multiple candidate text domains according to the N reference text domains.

[0093] As an optional implementation, the screening unit 23 is further used to: obtain the text domain length of each of the reference text domains; obtain the domain weight of each of the reference text domains according to the text domain type to which each of the reference text domains belongs; and screen the N reference text domains according to the text domain length and the domain weight to obtain the multiple candidate text domains.

[0094] As an optional implementation, the device further includes: a keyword truncation unit 25, which is used to perform keyword truncation on the tagged text domains in the tagged text domain set before screening the tagged text domains in the tagged text domain set, and remove the keywords in the tagged text domain that are ranked after the Mth position in descending order of word weight.

[0095] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0096] Figure 31 is a block diagram of an electronic device 800 for implementing picture information processing according to an exemplary embodiment. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0097] Reference Figure 3 , the electronic device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / display (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0098] The processing component 802 generally controls the overall operation of the electronic device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0099] The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0100] The power supply component 806 provides power to the various components of the electronic device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 800.

[0101] The multimedia component 808 includes a screen that provides a presentation interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0102] The audio component 810 is configured to present and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the electronic device 800 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for presenting audio signals.

[0103] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: home button, volume button, start button, and lock button.

[0104] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the electronic device 800. For example, the sensor assembly 814 can detect the open / closed state of the device 800, the relative positioning of components, such as the display and keypad of the electronic device 800, and the sensor assembly 814 can also detect the position change of the electronic device 800 or a component of the electronic device 800, the presence or absence of contact between the user and the electronic device 800, the orientation or acceleration / deceleration of the electronic device 800, and the temperature change of the electronic device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0105] The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0106] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above methods.

[0107] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by a processor 820 of an electronic device 800 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0108] A non-temporary computer-readable storage medium, when the instructions in the storage medium are executed by the processor of a mobile terminal, enables the mobile terminal to execute a method for processing image information, the method comprising: performing repeated image clustering on images in a web page, obtaining cluster images of each cluster and a set of tagged text domains of the cluster images; for each cluster image, obtaining keywords contained in each tagged text domain in the tagged text domain set and the word weights of the keywords according to the tagged text domain set; screening the tagged text domains in the tagged text domain set according to the word weights of the keywords in the tagged text domains and the word frequencies of the keywords in the corresponding tagged text domains to obtain multiple candidate text domains; removing similar texts from the multiple candidate text domains, and obtaining descriptive text information of the cluster images according to the multiple candidate text domains after removing similar texts.

[0109] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this disclosure. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

[0110] It should be understood that the present invention is not limited to the precise structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for processing image information, characterized in that: The method comprises: Repeatedly clustering the images in the webpage to obtain cluster images of each cluster and a set of marked text fields of the cluster images; For each cluster image, obtaining keywords and word weights of the keywords contained in each tagged text domain in the tagged text domain set according to the tagged text domain set; According to the word weight of the keyword in the marked text domain and the word frequency of the keyword in the corresponding marked text domain, the marked text domain in the marked text domain set is screened to obtain a plurality of candidate text domains; Removing similar texts from the multiple candidate text fields, and obtaining descriptive text information of the cluster image based on the multiple candidate text fields after removing similar texts; The step of screening the marked text domains in the marked text domain set according to the word weights of the keywords in the marked text domains and the word frequencies of the keywords in the corresponding marked text domains to obtain a plurality of candidate text domains includes: Calculate and obtain the score of each marked text field according to the word weight of the keyword in the marked text field and the word frequency of the keyword in the corresponding marked text field; Obtaining the first N marked text fields with the largest scores in the marked text field set as reference text fields; Obtain the plurality of candidate text domains according to N reference text domains; The step of obtaining the plurality of candidate text domains according to the N reference text domains includes: Obtaining the text field length of each of the reference text fields; According to the text domain type to which each of the reference text domains belongs, obtaining a domain weight of each of the reference text domains; The N reference text domains are screened according to the text domain length and the domain weight to obtain the multiple candidate text domains.

2. The method according to claim 1, characterized in that The step of obtaining the keywords and the word weights of the keywords contained in each of the marked text domains in the marked text domain set according to the marked text domain set includes: Obtaining keywords in each of the tag text fields; The following target parameters are obtained for each keyword: the word frequency and the number of occurrences of the keyword in the corresponding markup text domain set and the number of website domain names corresponding to the keyword; According to the target parameter of each keyword, the word weight of each keyword is calculated.

3. The method according to claim 2, characterized in that Calculating and obtaining the word weight of each keyword according to the target parameter of each keyword includes: For each of the keywords, according to the word frequency and the number of occurrences of the keyword in each of the markup text domains, the importance of the keyword in each of the markup text domains is calculated, and the importance is accumulated by decaying according to the word frequency and the number of occurrences; The word weight of each keyword is calculated based on the importance of the keyword in all the corresponding markup text domains and the number of website domain names corresponding to the keyword.

4. The method according to claim 1, characterized in that Before screening the mark text fields in the mark text field set, the method further includes: Keyword truncation is performed on the tag text fields in the tag text field set, and keywords after the Mth position in the tag text field that are sorted from large to small in terms of word weight are removed.

5. A device for processing image information, characterized in that: The device comprises: A clustering unit, used for clustering repeated images in the webpage, obtaining cluster images of each cluster and a set of marked text fields of the cluster images; A weight acquisition unit, used for acquiring, for each cluster image, keywords contained in each tagged text domain in the tagged text domain set and word weights of the keywords according to the tagged text domain set; A screening unit, configured to screen the marked text domains in the marked text domain set according to the word weights of the keywords in the marked text domains and the word frequencies of the keywords in the corresponding marked text domains, to obtain a plurality of candidate text domains; The screening unit is used to calculate the score of each marked text field according to the word weight of the keyword in the marked text field and the word frequency of the keyword in the corresponding marked text field; Obtaining the first N marked text fields with the largest scores in the marked text field set as reference text fields; Obtain the plurality of candidate text domains according to N reference text domains; The screening unit is further used to obtain the text field length of each of the reference text fields; According to the text domain type to which each of the reference text domains belongs, obtaining a domain weight of each of the reference text domains; Screening the N reference text domains according to the text domain length and the domain weight to obtain the multiple candidate text domains; The similarity removal unit is used to remove similar texts from the multiple candidate text domains, and obtain the description text information of the cluster image according to the multiple candidate text domains after the similar texts are removed.

6. The device according to claim 5, characterized in that The weight acquisition unit is used for: Obtaining keywords in each of the tag text fields; The following target parameters are obtained for each keyword: the word frequency and the number of occurrences of the keyword in the corresponding markup text domain set and the number of website domain names corresponding to the keyword; According to the target parameter of each keyword, the word weight of each keyword is calculated.

7. An electronic device, characterized in that: It includes a memory and one or more programs, wherein one or more of the programs are stored in the memory and are configured to be executed by one or more processors to execute the operation instructions corresponding to the method as described in any one of claims 1 to 4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps corresponding to the method according to any one of claims 1 to 4 are implemented.

Images