System for search and method for operating thereof

• US 9,858,297 B2
• Filed: 07/24/2014
• Issued: 01/02/2018
• Est. Priority Date: 10/25/2013
• Status: Active Grant

• Current Assignee: Hanwha Vision Company Limited
• Original Assignee: Hanwha Techwin Co., Ltd. (Hanwha Corporation)
• Inventors: Park, Dong Jun, Ryu, Yeon Geol, Shin, Hak Chul, Jung, Dong Whan
• Primary Examiner(s): Park, Chan
• Assistant Examiner(s): Thomas, Mia M
• Application Number:
  US14/340,114
• Publication Number:
  US 20150117759A1
• Time in Patent Office: 1,258 Days
• CPC Class Codes: G06F 16/51 Indexing; Data structures therefor; Storage structures

Abstract

Provided are a system for search and a method for operating thereof. The system for search includes: a preliminary data analysis part which extracts a variety of attributes through analysis of images being input, analyzes a trend about a category as information requested by a client with the image analysis result, and stores the trend analysis result as metadata; an index part which stores the image analysis result, and structuralizes, organizes and stores the stored metadata in order to easily search the metadata; and a search part which extracts trend information matching a category input by a client, from the index part and provides the trend information in a predetermined format.

Claims

1. A system for search comprising at least one processor to implement:

a preliminary data analysis part configured to analyze input images to extract a variety of attributes, analyze a trend about a category of information requested by a client by using the extracted variety of attributes, and store, based on trend analysis, trend information as metadata;

an index part configured to make the metadata searchable by structuralizing the metadata and storing the structuralized metadata; and

a search part configured to extract the trend information matching the category of information requested by the client from the searchable metadata and provide the trend information in a predetermined format,

wherein the preliminary data analysis part comprises;

an occurrence frequency analysis unit configured to analyze, as first trend information of the trend information, an occurrence frequency of the category in a random time period;

a change analysis unit configured to analyze, as second trend information of the trend information, a change in the first trend information with respect to a set time; and

a movement line analysis unit configured to provide, as third trend information of the trend information, a result of movement line analysis of the category, through detection and tracking of a motion in the category.

2. The system of claim 1, wherein the preliminary data analysis part further comprises:

an image analysis engine configured to analyze the input images and extract the variety of attributes; and

a first storage unit configured to store the trend information as the metadata.

3. The system of claim 2, wherein the index part comprises:

a second storage unit configured to store an output of the image analysis engine;

an indexing engine configured to structuralize and organize metadata stored in the first storage unit to make the metadata searchable; and

a third storage unit configured to store the searchable metadata from the indexing engine.

4. The system of claim 3, wherein the search part comprises:

a query engine configured to classify a search word input from the client into a category;

a search engine configured to search the second and third storage units for trend information corresponding to the classified category; and

a browsing engine configured to provide a search result of the trend information by the search engine in the predetermined format.

5. The system of claim 1, wherein the search part is configured to provide metadata of the first trend information, analyzed by the occurrence frequency analysis unit, in an output screen format that includes upper categories indicated on one axis and, on another axis, visual graphics representing frequencies of occurrences of lower categories included in each upper category.

6. The system of claim 5, wherein, in response to a random lower category being selected from among the lower categories, detailed information of the selected lower category is displayed.

7. The system of claim 1, wherein the search part is configured to provide metadata of the second trend information, in a heat-map format in which a change in occurrence frequencies with respect to the set time is expressed in visual graphics of a heat distribution form.

8. The system of claim 1, wherein the search part is configured to provide metadata of the second trend information by changing the first trend information, analyzed by the occurrence frequency analysis unit, to the second trend information, in a graph format in which a change in occurrence frequencies with respect to the set time is expressed in visual graphics.

9. The system of claim 1, wherein the movement line analysis unit comprises:

a movement line information generation unit configured to generate movement line information through motion detection and tracking in relation to the category from the input images;

a noise removing unit configured to remove noise of the movement line information;

a clustering unit configured to make clusters each of which is formed of similar movement line information items in which the noise is removed; and

a grouping unit configured to provide, as the third trend information, a result of grouping similar movement line information items by performing artificial intelligence neural network learning with the clustered movement line information items.

10. The system of claim 9, wherein the search part is configured to provide metadata of the third trend information, which is retrieved from the index part, in an output screen format which is formed with a first area where a movement line in an image is displayed and a second area where the result of grouping similar movement line information items by performing artificial intelligence neural network learning is displayed as a self-organizing map.

11. The system of claim 10, wherein, in response to any one group being selected in the second area, a movement line corresponding to the selected group is displayed in the first area.

12. The system of claim 1, wherein the movement line analysis unit comprises:

a movement line information generation unit configured to generate movement line information through motion detection and tracking in relation to the category from the input images;

a noise removing unit configured to remove noise of the movement line information;

a point selection unit configured to select arbitrary points included in the movement line information in which the noise is removed;

a clustering unit configured to make clusters each of which is formed of similar movement line information items including the points; and

a movement probability calculation unit configured to calculate a probability of moving from any one clustered point to another clustered point, and provide the probability as the third trend information.

13. The system of claim 12, wherein metadata of the third trend information retrieved from the index part is provided in an output screen format in which clustered points in an image are displayed and the probability of moving from any one clustered point to the another clustered point is shown as an arrow.

14. The system of claim 13, a magnitude of the probability of moving is proportional to a size of the arrow.

15. A method for operating a system for search, the method comprising:

extracting, for preliminary data analysis, a variety of attributes by analyzing input images, analyzing a trend about a category of information requested by a client by using the extracted variety of attributes, and storing, based on trend analysis, trend information as metadata;

indexing the stored metadata by structuralizing the metadata and storing the structured metadata to make the metadata searchable; and

extracting, from the searchable metadata, the trend information matching a category input by the client, and providing the extracted trend information in a predetermined format,

wherein the analyzing the trend about the category comprises;

analyzing, as first trend information of the trend information, an occurrence frequency of the category in a random time period;

analyzing, as second trend information of the trend information, a change in the first trend information with respect to a set time, and

providing, as third trend information of the trend information, a result of movement line analysis of the category, through detection and tracking of a motion in the category.

16. The method of claim 15, wherein the extracting the variety of attributes comprises extracting the variety of attributes through analysis of the input images, and

wherein the analyzing the trend about the category further comprises, in response to the variety of attributes being extracted, analyzing the trend about the category and providing the trend analysis as the trend information.

17. The method of claim 15, wherein the providing of the result of movement line analysis comprises:

generating movement line information through motion detection and tracking in relation to the category from the input images;

removing noise of the movement line information;

making clusters each of which is formed of similar movement line information items in which the noise is removed; and

providing, as the third trend information, a result of grouping similar movement line information items by performing artificial intelligence neural network learning with the clustered movement line information items.

18. The method of claim 15, wherein the providing of the result of movement line analysis comprises:

generating movement line information through motion detection and tracking in relation to the category from the input images;

removing noise of the movement line information;

selecting arbitrary points included in the movement line information in which the noise is removed;

making clusters each of which is formed of similar movement line information items including the points; and

calculating a probability of moving from any one clustered point to another clustered point, and providing the probability as the third trend information.