Processing

Please wait...

Settings

Settings

Goto Application

1. WO2012058794 - IMAGE SEARCH

Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

[ EN ]

Image Search

BACKGROUND

[oooi] Users may interact with a search engine to obtain search results for a wide variety of content. However, traditional techniques that were utilized to provide search results may become limited when confronted with different types of content. For example, traditional search results included links to websites with brief textual descriptions. These search results were typically provided in response to keywords in a search request. However, the functionality of these traditional techniques may be limited in relation to other types of content, such as images.

SUMMARY

[0002] Image search techniques are described. In one or more implementations, images in a search result are ordered based at least in part on similarity of the images, one to another. The search result having the ordered images is provided in response to a search request.

[0003] In one or more implementations, a search request is received having at least one term that is indicative of at least two or more different types of images. A search result is provided in response to the search request. The search result includes a first type of the image positioned closer to a beginning of the search result than a second type of the image. The first type of the image has a larger number of similar images than the second type.

[0004] In one or more implementations, a plurality of binary words is generated, each to describe content in respective one of a plurality of images. The plurality of binary words are compared, one or another, to determine similarity of respective images. Clusters are formed based on the comparison of the determined similarity of the images. The images are ranked based on a number of similar images in a respective cluster for inclusion in a search result to be provided in response to a search query received via an internet.

[0005] This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The detailed description is described with reference to the accompanying figures. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. The use of the same reference numbers in different instances in the description and the figures may indicate similar or identical items.

[0007] FIG. 1 is an illustration of an environment in an example implementation that is operable to perform a search.

[0008] FIG. 2 is an illustration of a system in an example implementation showing a search module of FIG. 1 as processing images to obtain a search result for provision to a client.

[0009] FIG. 3 is a flow diagram depicting a procedure in an example implementation in which a search result is configured for a search request that is indicative of two or more different types of content in the images.

[0010] FIG. 4 is a flow diagram depicting a procedure in an example implementation in which images are ordered in a search result based on similarity, [ooii] FIG. 5 is a flow diagram depicting a procedure in an example implementation in which a binary word is generated to describe content in images and used as a basis of comparison to determine similarity of the images, one to another.

DETAILED DESCRIPTION

Overview

[0012] Traditional techniques to perform a search may be limited when confronted with different types of content. For example, traditional techniques that are utilized to search for a webpage may provide a search result that is difficult to decipher when requested to search for a particular image.

[0013] Image search techniques are described. In one or more implementations, search techniques are employed that may help to differentiate between different types of images. For example, a search query may be received that may relate to a different person, place, or thing. Traditional image search techniques, when encountering such a situation may intermix the results such that a user may have a difficult time in locating a particular image of interest. However, in this example, clustering may be employed to form groups of images based on similarity. These groups may then serve as a basis for ordering a search result, thereby increasing the likelihood that a user may locate a particular image of interest. Further discussion of these and other techniques may be found in relation to the following sections.

[0014] In the following discussion, an example environment is first described that is operable to perform image search techniques described herein. Example procedures are then described, which are operable in the example environment as well as in other environments. Likewise, the example environment is not limited to performance of the example procedures. Further, although the search techniques are described in relation to search for images, these techniques may be employed to search for a variety of different types of data, such as webpages, media, documents, files, and so on.

Example Environment

[0015] FIG. 1 is an illustration of an environment 100 in an example implementation that is operable to employ image search techniques. The illustrated environment 100 includes a search service 102, a client 104, and an image provider 106, each of which are communicatively coupled, one to another, over a network 108.

[0016] Although the client 104 is illustrated as a client device (e.g., a traditional desktop computer) and the search service 102 and image provider 106 are illustrated as being implemented by one or more servers, these entities may be implemented by a variety of different devices. For example, the client 104 may be configured as a computer that is capable of communicating over the network 108, such as a desktop computer, a mobile station, an entertainment appliance, a set-top box communicatively coupled to a display device, a wireless phone, a game console, a tablet computer, a netbook, and so forth. Thus, the client 104 may range from a full resource device with substantial memory and processor resources (e.g., personal computers, game consoles) to a low-resource device with limited memory and/or processing resources (e.g., traditional set-top boxes, hand-held game consoles). The client 104 may also describe logical clients that include software and/or as well as hardware that is used to execute the software, e.g., one or more processors, functional blocks, and so on.

[0017] Thus, the client may describe a computing device that may also include an entity (e.g., software) that causes hardware of the computing device to perform operations, e.g., configures processors, functional blocks, and so on. For example, the computing device may include a computer-readable medium that may be configured to maintain instructions that cause the computing device, and more particularly hardware of the computing device to perform operations. Thus, the instructions function to configure the hardware to perform the operations and in this way result in transformation of the hardware to perform the operations. The instructions may be provided by the computer-readable medium to the computing device 102 through a variety of different configurations.

[0018] One such configuration of a computer-readable medium is signal bearing medium and thus is configured to transmit the instructions (e.g., as a carrier wave) to the hardware of the computing device, such as via the network 104. The computer-readable medium may also be configured as a computer-readable storage medium and thus is not a signal bearing medium. Examples of a computer-readable storage medium include a random-access memory (RAM), read-only memory (ROM), optical discs, flash memory, hard disk memory, and other memory devices that may use magnetic, optical, and other techniques to store instructions and other data.

[0019] Although the network 108 is illustrated as the Internet, the network may assume a wide variety of configurations. For example, the network 108 may include a wide area network (WAN), a local area network (LAN), a wireless network, a public telephone network, an intranet, and so on. Further, although a single network 108 is shown, the network 108 may be configured to include multiple networks.

[0020] The search service 102 is illustrated as including a search module 110. The search module 110 is representative of functionality to provide a search result 1 12 in response to a search query. For example, a user of the client 104 may interact with a communication module 114, which is representative of functionality of the client 104 to interact with the network 108, such as a browser over the Internet.

[0021] The search module 110 may employ a variety of different techniques to form a search result 1 12. For example, the search module 110 may employ one or more software robots ("bots") to search and index content available over the Internet. These indexes may be based on keywords and other information. Further, a variety of different techniques may be employed to apply different weights to parts of the index such that a user has an increased likelihood of finding content of interest.

[0022] For a given query for images, however, it may difficult to find a representative set of images that best represent the query using traditional techniques. For example, the search module 110 may index images 116 from a plurality of different image providers 106. However, in some instances a single search query may relate to different types of content in the image, such as a different person, place, or thing. Accordingly, the search module 110 in this example may employ techniques that may be used to increase a likelihood of a

user receiving a search result 1 12 that includes a desired image, further discussion of which may be found in relation to FIG. 2.

[0023] Although an example of Internet search was described in relation to FIG. 1 , these techniques may be leveraged for a variety of other uses. For example, the client 104 may also employ a search module 1 18 to search images 120 that are local to the client 104, images 1 16 accessible remotely via the network 108, and so on. A variety of other examples are also contemplated, and accordingly this discussion is not limited to the example environment 100.

[0024] Generally, any of the functions described herein can be implemented using software, firmware, hardware (e.g., fixed logic circuitry), manual processing, or a combination of these implementations. The terms "module," "functionality," and "logic" as used herein generally represent software, firmware, hardware, or a combination thereof. In the case of a software implementation, the module, functionality, or logic represents program code that performs specified tasks when executed on a processor (e.g., CPU or CPUs). The program code can be stored in one or more computer readable memory devices. The features of the image search techniques described below are platform-independent, meaning that the techniques may be implemented on a variety of commercial computing platforms having a variety of processors.

[0025] FIG. 2 is an illustration of a system 200 in an example implementation showing the search module 110 of FIG. 1 as processing images to obtain a search result. The search module 110 in this example accesses images 1 16 that are

available via the network 108. The search module 1 10 then employs a plurality of buckets 202, 204, 206, 208 to cluster images together based on similarity of the images, one to another, as computed using a similarity module 210. This similarity may thus serve as a basis to provide a search result 1 12 that has an increased likelihood of providing a user an opportunity to locate an image 116 of interest.

[0026] For example, the search module 1 10 may be employed to process images for a keyword, which in this instance is "Seattle." Accordingly, the search module 110 may scan the images 116 for similarities using the similarity module 210. For instance, the similarity module 210 may examine the images to determine what content is included in the images. The similarity module 210 may then assign an identifier to this content and compare the identifiers to determine which images are similar. A variety of different techniques may be used to determine similarity, an example of which is described in relation to FIG. 5.

[0027] Similar images may then be formed into clusters. Continuing with the previous example, images for the keyword "Seattle" that contain content relating to the Space Needle may be assigned to a first bucket 202. Likewise, images for the keyword "Seattle" that contain content relating to Mount Rainier may be assigned to a second bucket 204. Further, images for the keyword "Seattle" that contain content relating to the Seattle skyline may be assigned to a third bucket 206. This process may continue for a variety of different content, including images of a family vacation to Seattle as illustrated for the fourth bucket 208. [0028] A number of images assigned to the different buckets 202-208 may then serve as a basis for ordering a search result. In the illustrated example, the buckets 202-208 are arranged in an ordered indicating a number of images that were assigned to the respective bucket, such that bucket 202 was assigned more images than bucket 204, and so on.

[0029] This ordering may then be used as a basis to arrange images in a search result, such that images from the first bucket 202 of the Space Needle may be moved up in rank in the search result 1 12. In this way, a user is more likely to obtain an image of interest in response to a search query. These techniques may be leveraged to provide a variety of other functionality, further discussion of which may be found in relation to the following procedures.

Example Procedures

[0030] The following discussion describes image search techniques that may be implemented utilizing the previously described systems and devices. Aspects of each of the procedures may be implemented in hardware, firmware, or software, or a combination thereof. The procedures are shown as a set of blocks that specify operations performed by one or more devices and are not necessarily limited to the orders shown for performing the operations by the respective blocks. In portions of the following discussion, reference will be made to the environment 100 of FIG. 1 and the system 200 of FIG. 2.

[0031] FIG 3 depicts a procedure 300 in an example implementation in which a search result is configured for a search request that is indicative of two or more different types of content in the images. A request is received having at least one term that is indicative of at least two or more different types of content in the images (block 302). For example, the request may be received having a search term "jaguar." However, the term jaguar may relate to an automobile as well as an animal. Previous techniques that were used to provide images often resulted in a confused result in which both types of images were intermixed. Using the techniques described herein, though, the search result may differentiate between the different types.

[0032] Accordingly, in this example a search result is provided in response to the search request, the search result having images that include a first type of the content positioned closer to a beginning of the search result than images having a second type of the content, the first type of the image having a larger number of similar images than the second type (block 304). Continuing with the previous example, the search module 110 may employ techniques to differentiate between images of the automobile and the animal for "jaguar."

[0033] For instance, the search module 110 may be configured to promote similar images that are more common. In this way, images have a more common content in a meaning for the term (e.g., jaguar) may be promoted in a ranking of the search result 1 12 over less common meanings. A variety of different techniques may be employed to determine similarity, further discussion of which may be found in relation to the following figure. Although in this example a request was described as being received and search results were then provided in response to the request, it should be readily apparent that the search results may be computed in a variety of ways. For example, the search results 112 may be pre-computed before receipt of the request (e.g., using indexing techniques described in relation to FIG. 2), in response to the request, and so on.

[0034] FIG. 4 depicts a procedure 400 in an example implementation in which images are ordered in a search result based on similarity. Images in a search result are ordered based at least in part on similarity of the images, one to another (block 402). For example, similar images may be grouped into image clusters (block 404). This similarity may be determined in a variety of ways, such as through an examination of the content of the images including the use of recognition techniques that may be used to identify a face or other object. An example of one such technique using binary words to examine content in images is described in relation to FIG. 5.

[0035] A first image cluster that contains a higher number of grouped similar images is ranked higher in a search result than a second image cluster that contains a lesser number of grouped similar imaged (block 406). Thus, a number of images in respective clusters may be used to drive "where" the images are included in a search result 112, such by ranking them higher or lower depending on pervasiveness of content included in the images and similarity of that content across those images. Thus, the search result 112 is more likely to include images having content that are of interest based on the interest shown in that content by the number of those images that are available.

[0036] FIG. 5 depicts a procedure 500 in an example implementation in which a binary word is generated to describe content in images and used as a basis of comparison to determine similarity of the images, one to another. A plurality of binary words is generated, each to describe content in a respective one of a plurality of images (block 502). For example, the search module 1 10 may examine content in images and describe the content through use of a binary word. The binary word, for instance, may describe an automobile included in the image, an animal included in the image, and so on. In an implementation, an image may have a plurality of corresponding binary words that may be used to identify content in the image, such as a car and logo contained in the image.

[0037] The plurality of binary words are compared, one to another, to determine similarity of respective images (block 504). Continuing with the previous example, the binary words may be compared to each other by the search module 110. In this way, the search module 110 may perform an efficient comparison of a multitude of images to determine similarity.

[0038] Clusters are formed based on the comparison of the determined similarity of the images (block 506). As described in relation to FIG. 2, the search module 110 may form the clusters by assigning images to buckets based on similarity of the images. Thus, images in a particular bucket are determined to be similar and may give an accurate comparison of a number of this type of image in relation to a number of other types of images assigned to other buckets.

[0039] The images are ranked based on a number of similar images in a respective cluster for inclusion in a search result to be provided in response to a search query received via the internet (block 508). Continuing with the previous example, the clusters formed by the "bucketing" of the images may be used to give an idea of relative prevalence of content in the images with respect to each other. Accordingly, the search module 110 may leverage this knowledge to order the search result 1 12 to follow this prevalence, thereby increasing a likelihood that a user will be presented with an image of interest. Further, in an implementation the search results may be clustered such that each cluster is represented by the similar image in the search result 1 12. In this way, a user may quickly navigate through different images to locate a particular one of interest without viewing a same of very similar image multiple times. In an implementation, this representative image may be selectable to cause output of additional images from the cluster.

Conclusion

[0040] Although the invention has been described in language specific to structural features and/or methodological acts, it is to be understood that the invention defined in the appended claims is not necessarily limited to the specific features or acts described. Rather, the specific features and acts are disclosed as example forms of implementing the claimed invention.