Method for searching multimedia using progressive histogram

• US 20020059206A1
• Filed: 07/20/2001
• Published: 05/16/2002
• Est. Priority Date: 07/21/2000
• Status: Active Grant

• Current Assignee: LG Electronics, Inc. (LG Corporation)
• Original Assignee: LG Electronics, Inc. (LG Corporation)
• Inventors: Kim, Heon-Jun, Song, Jung Min, Lee, Jin-Soo
• Granted Patent: US 7,039,255 B2
• Time in Patent Office: Days
• US Class Current: 707/3
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation

Abstract

The present invention relates to a method for retrieving multimedia using a histogram. The present invention configures a histogram that a relatively important bin comes at the lead of the access when accessing histogram data for retrieving the multimedia and retrieves multimedia data by using the progressive histogram. According to the present invention, there are several advantages that a high retrieval capacity can be guaranteed even though only front part in the order of transfer of the histogram is used and the efficiency and capacity of retrieval can be enhanced by using a progressive histogram optimized according to the feature at every multimedia.

Claims

1. A histogram in which a relatively important bin comes at the lead of the access on the basis of the access order of time series grounds in order to retrieve multimedia using the histogram.

2. The histogram according to claim 1, wherein the histogram is a color histogram.

4. The progressive multimedia retrieval method in accordance with claim 2, wherein the comparison and retrieval of the query histogram and the retrieval target histogram are performed as information having the meaning that the progressive part, which comes at the lead relatively, represents the whole corresponding histogram, when accessing the histogram reconfigured as the order of importance,

5. The progressive multimedia retrieval method in accordance with claim 2, wherein the reconfiguration of the histogram in the order of importance of bins is considered that bins having a larger variance are more important bin by means of calculation based on the variance of bin values for each bin calculated from a sample group.

6. The progressive multimedia retrieval method in accordance with claim 2, wherein the reconfiguration of the histogram in the order of importance of bins is considered that bins having a smaller variance in its group but having a larger variance between other groups are more important bins by means of calculation of the variance of each bin in its group and the variance of each bin between other groups from the sample group which is classified into groups by similarity of images.

7. The progressive multimedia retrieval method in accordance with claim 2, wherein the reconfiguration of the histogram in the order of importance of bins is configured by sampling bins meaning the segmental regions segmented by quantization of a color space in which a segmental region having a low fineness is selected with high probability and appeared in the front part of the histogram.

8. The progressive multimedia retrieval method in accordance with claim 6, wherein, when the histogram uses a color space of HMMD configured by Hue, SUM(MAX(RGB)+MIN(RGB)), DIFF(MAX(RGB)−

MIN(RGB)) and the HMMD color space has the range of DIFF values of 0˜

255 and the designated DIFF values for segmenting on the basis of the shaft of DIFF are 9,29,75,200 and the five segmental regions segmented on the basis of the DIFF values are SI,S2,S3,S4 and S5,respectively, the five segmental regions are configured as 184 bins by using a color quantization method in which S1 is segmented by 8 on the basis of SUM, S2 is segmented by 4 on the basis of SUM and by 8 on the basis of Hue, S3 is segmented by 4 on the basis of SUM and by 12 on the basis of Hue, S4 is segmented by 4 on the basis of SUM and by 12 on the basis of Hue, and S5 is segmented by 2 on the basis of SUM and by 24 on the basis of Hue.

9. The progressive multimedia retrieval method in accordance with claim 8, wherein the sampling probabilities at the regions of S1,S2,S3,S4 and S5, respectively, are decided on 24:

12;

6;

2;

1 in order to make the sampling probabilities differ.

10. The progressive multimedia retrieval method in accordance with claim 2, wherein, in the multimedia retrieval system using the progressive histogram, the method further comprises the step of:

transferring the histogram to a client or retrieving the histogram by using the histogram in part or whole from the front part of the histogram according to a retrieval object and a hardware environment of the client for the retrieval.

11. The progressive multimedia retrieval method in accordance with claim 2, wherein, in the multimedia retrieval system using the progressive histogram, the method further comprises retrieving by only using the received histogram data, when the histogram data of the query data is interrupted by unexpected results in transfer.

12. The progressive multimedia retrieval method in accordance with claim 2, wherein the histogram is a color histogram.

14. The progressive multimedia retrieval method in accordance with claim 13, wherein the histogram is a color histogram.

16. The progressive multimedia retrieval method in accordance with claim 15, wherein the histogram is a color histogram.

17. The multimedia retrieval method using a progressive histogram in accordance with claim 15, wherein, in the multimedia retrieval system using the progressive histogram, the method further comprises transferring the histogram to a client or retrieving the histogram by using the histogram in part or whole from the front part of the histogram according to a retrieval object and a hardware environment of the client for the retrieval.

18. The multimedia retrieval method using a progressive histogram in accordance with claim 15, wherein, in the multimedia retrieval system using the progressive histogram, the method further comprises the steps of:

reconfiguring a query histogram and a retrieval target histogram according to the respective bin order information; and

comparing/retrieving the histogram reconfigured that an important bin comes at the lead according to the bin order information.

19. The multimedia retrieval method using a progressive histogram in accordance with claim 15, wherein, in the multimedia retrieval system using the progressive histogram, the method further comprises the steps of:

selecting a query histogram and a retrieval target histogram one by one according to the bin order information; and

comparing/retrieving the histogram according to the bin order.

3. A progressive multimedia retrieval method comprising the steps of;

reconfiguring a histogram that a relatively important bin comes at the lead of the access when accessing histogram data for retrieving; and

comparing/retrieving a query histogram and a retrieval target histogram according to the order of importance of bins, when accessing and then comparing/retrieving the histogram reconfigured as the order of importance, the access order reconfigured as the order of importance.

13. A histogram information representation structure comprising:

bin order information deciding an order that a relatively important bin comes at the lead of the access in the access order of time series grounds in order to query using the histogram.

15. A multimedia retrieval method using a progressive histogram comprising the steps of:

producing bin order information deciding an access order that a relatively important bin comes at the lead of the access when accessing histogram data for retrieving in the multimedia retrieval using the progressive histogram;

retrieving partial bin of the histogram by using the bin order information; and

comparing/retrieving the histogram by selecting the bin according to the order of importance of bins.

20. A multimedia retrieval method using a progressive histogram comprising the steps of:

calculating importance of bins configuring the corresponding histogram in each multimedia object, in which the importance is differ according to the viewpoints of comparing/retrieving, when calculating the importance of bins configuring histogram as feature information for retrieving multimedia; and

comparing/retrieving the histogram according to the order of importance by using the calculated information of importance according to the viewpoints of comparing/retrieving.

21. The multimedia retrieval method using a progressive histogram in accordance with claim 20, the method further comprises the step of:

calculating the information of importance of bins according to the viewpoints of comparing/retrieving on the basis of user'"'"'s feedback information to similarity or dissimilarity or on the basis of group information to a similarity object obtained in advance.

22. The multimedia retrieval method using a progressive histogram in accordance with claim 20, wherein the histogram is transmitted by the order of importance of bins and the partial histogram or the whole histogram is transmitted according to the situation of a system or a network.

23. The multimedia retrieval method using a progressive histogram in accordance with claim 20, wherein the histogram is a color histogram.

25. The method for configuring progressive histogram information in accordance with claim 24, wherein the order information of bin of the histogram is a one-dimensional matrix type arranged the order of each bin of histogram.

26. The method for configuring progressive histogram information in accordance with claim 24, wherein the histogram is a color histogram.

24. A method for configuring progressive histogram information comprising the steps of:

producing importance of bins configuring the histogram as feature information for retrieving multimedia by considering the feature according to the viewpoints of comparing/retrieving, in which the importance is differ in each multimedia object; and

producing and storing the order information of bin of the histogram.