| CPC G06V 30/413 (2022.01) [G06V 30/245 (2022.01); G06V 30/416 (2022.01)] | 9 Claims |
|
1. An information processing apparatus for dividing a plurality of page data that is scanned, comprising:
an OCR unit that performs optical character recognition for character and layout in a page for each of the plurality of page data;
a rule order unit configured to classify each of the plurality of page data based on a page ordering rule according to the character and the layout that are recognized by the performance of optical character recognition by the OCR unit and divide the plurality of page data into page units; and
a rule setting unit configured to:
classify each of the plurality of page data by using the layout and font in a page and set the rule by extracting page number based on classified coordinates and character string in a page;
extract a character string indicating a page number from a corresponding page of the plurality of page data according to the layout in a classified character layout data;
when the character string indicating the page number is extracted, acquire coordinates and font of an area where the character string indicating the page number is located; and
set a rule for extracting a page number based on the coordinates and font of the page number area in association with the classified layout data;
calculate a certainty of page extraction (‘rule page certainty’) by comparing a set layout to the layout of a page from which the page number was extracted based on the rule for extracting page number;
wherein the rule order unit is configured to calculate division certainty by similarity of the layout, similarity of font, and the similarity of extraction result of the page number;
wherein the information processing apparatus further comprises:
an ML order unit configured to classify page data of a page with low division certainty calculated by the rule order unit by machine learning and perform division;
wherein the ML order unit is configured to perform classification by any one or any combination of
a model by using the layout and image data of the page,
a model by using the layout and characters, and
a model for detecting abnormal value and the certainty of classification result by the rule order unit,
wherein the ML order unit is configured to infer the division position based on a model by using the layout and characters;
wherein the ML order unit is configured to infer a page number of a page by a machine learning model and calculating a certainty of the inference of page number (‘ML page certainty’);
wherein the information processing apparatus further comprises an entire certainty calculation unit configured to calculate entire division certainty from the division certainty calculated by the rule order unit and inference certainty of the division position extracted by the ML order unit; and
wherein the entire division certainty is an average of the division certainty for each page calculated by equation (1):
(Certainty for each page)={(Larger value of the rule page certainty and the ML page certainty)+(Larger value of the division certainty and the inference certainty)}/2.
|