Apparatus for judging forgery webtoon contents using extracting region and method thereof
An apparatus for judging forgery webtoon contents using extraction region, includes an original webtoon feature extraction unit that extracts the webtoon in cut units, divides the webtoon into a plurality of regions, and extracts feature information for each of a plurality of divided regions when original content is input; a query target webtoon feature extraction unit that extracts the webtoon in cut units, removes an unnecessary space or a text region, divides the webtoon into a plurality of regions, extracts feature information for each of the plurality of divided regions, and selects the plurality of regions when query target content is input; and a determination unit that compares feature information of original data regions that match the selected plurality of query data regions to determine a similarity, and determines whether the query target content is forged and altered using a determination result.
1 . An apparatus for judging forgery webtoon contents, the apparatus comprising:
an original webtoon feature extraction unit configured to:
extract a plurality of first cuts from an original content;
divide each of the first cuts into a plurality of first regions; and
extract feature information for each of the first regions;
a target webtoon feature extraction unit configured to:
extract a plurality of second cuts from a target content;
divide each of the second cuts into a plurality of second regions;
extract feature information for each of the second regions; and
select a predetermined number of the second regions based on the extracted feature information of the second regions; and
a determination unit configured to compare the feature information of the first regions that match the selected second regions with the feature information of the selected second regions to determine a similarity, and determine whether the target content is forged and altered using a determination result,
wherein the original webtoon feature extraction unit is further configured to remove unnecessary spaces and text regions in each of the first cuts, and the target webtoon feature extraction unit is further configured to remove unnecessary spaces and text regions in each of the second cuts, and
wherein the unnecessary spaces include a margin around an edge of each of the first and second cuts and a frame area surrounding each of the first and second cuts, and
the text regions include text inside speech ballons in each of the first and second cuts.
2 . The apparatus for judging forgery webtoon contents of claim 1 , wherein the original webtoon feature extraction unit or the target webtoon feature extraction unit is configured to:
cause an outline for each of the first or second cuts from which the unnecessary spaces and the text regions are removed to be sharpened by using a sharpening filter;
apply a binary filter to extract outline information;
extract the outline using a Canny Edge Detector; and
apply Threshold Binarization to extract feature information from the outline, and additionally apply a Speed Up Robust Feature (SURF) algorithm to extract the feature information of each of the first or second regions at a high speed.
3 . The apparatus for judging forgery webtoon contents of claim 1 , wherein the target webtoon feature extraction unit configured to select five second regions having the largest number of feature information among the plurality of second regions.
4 . The apparatus for judging forgery webtoon contents of claim 3 , wherein the determination unit configured to determine that the target content is forged and altered when there are three or more second regions having a similarity of 80% or more to the corresponding first regions among the selected five second regions.
5 . A method for judging forgery webtoon contents by using an apparatus for judging forgery webtoon contents, the method comprising:
extracting feature information of an original webtoon, the extracting of the feature information of the original webtoon includes:
extracting a plurality of first cuts from an original content;
dividing each of the first cuts into a plurality of first regions; and
extracting feature information for each of the first regions;
extracting feature information of a target webtoon, the extracting of the feature information of the target webtoon includes:
extracting a plurality of second cuts from a target content;
dividing each of the second cuts into a plurality of second regions;
extracting feature information for each of the second regions; and
selecting a predetermined number of the second regions based on the extracted feature information of the second regions; and
determining forgery of the target content using the feature information of the first and second regions, the determining of the forgery of the target content includes comparing the feature information of the first regions that match the selected second region with the feature information of the selected regions to determine a similarity, and determining whether the target content is forged and altered using a determination result,
wherein the extracting of the feature information of the original webtoon further includes removing unnecessary spaces and text regions in each of the first cuts, and the extracting of the feature information of the target webtoon further includes removing unnecessary spaces and text regions in each of the second cuts, and
wherein the unnecessary spaces include a margin around an edge of each of the first and second cuts and a frame area surrounding each of the first and second cuts, and
the text regions include text inside speech ballons in each of the first and second cuts.
6 . The method for judging forgery webtoon contents of claim 5 , wherein the of extracting of the feature information of the original webtoon or the target webtoon includes:
sharpening an outline for each of the first or second cuts from which the unnecessary spaces and the text regions are removed by using a sharpening filter;
applying a binary filter to extract outline information;
extracting the outline using a Canny Edge Detector;
applying Threshold Binarization to extract feature information from the outline; and
additionally applying a Speed Up Robust Feature (SURF) algorithm to extract the feature information of each of the first or second regions at a high speed.
7 . The method for judging forgery webtoon contents of claim 5 , wherein the selecting of the predetermined number of the second regions includes selecting five second regions having the largest number of feature information among the plurality of second regions.
8 . The method for judging forgery webtoon contents of claim 7 , wherein the determining of the forgery of the target content includes determining that the target content is forged and altered when there are three or more second regions having a similarity of 80% or more to the corresponding first regions among the selected five second reason.