Digital book layout analysis method

Abstract

The invention discloses a digital book layout analysis method. The digital book layout analysis method based on region segmentation is essentially a method for storing image segmentation and region classification in a JEPG or TIF format. The method is characterized by, to begin with, extracting image edges and carrying out over-segmentation region basic merging through a morphological algorithm and with a Gabor linear filter being combined to realize segmentation of a book region; then, carrying out region filling and ordering on segmented image blocks to reconstruct a reading order; and finally, carrying out feature extraction on an image region, and training a classifier to obtain attributes of each region, thereby realizing book region separation and improving identification correctness of an OCR engine and accuracy of book query.

Description

technical field

[0001] The invention relates to a layout analysis technology for pictures existing in image form in a digital library, in particular to a layout analysis technology for region division and classification. Background technique

[0002] With the research and development of computer and network technology, digital libraries are gradually developing from information-based processing and simple human-machine interface to knowledge-based processing and extensive understanding between machines, so that people can use computers and networks The ability to expand intellectual activities on a larger scale plays an extremely important role in all areas that need to communicate, disseminate, store, and utilize knowledge, including e-commerce, education, and telemedicine.

[0003] Since the books in the CADAL digital library exist in the form of images, they need to be processed by OCR to recognize the text and analyze the layout before they can provide in-depth services....

Examples

Embodiment

[0199] Below in conjunction with the method of the present invention, describe in detail the concrete steps that this example implements, take a certain page in the electronic book that CADAL library digital library scans as example here, as Figure 5 show, explain image 3 all the processes.

[0200] 1) Read the original image, perform grayscale processing on the color image, and read in the grayscale image

[0201] 2) Extract the edge of the picture. Create an 8*8 square structuring element to dilate the image. The image background is estimated by using the morphological opening operation imopen and closing operation imclose. The opening operation is to smooth the image contour, disconnect the narrow connection so as to realize the deletion of small protrusions, and construct a disc structure with a radius of 5 during the opening operation. The closing operation is equivalent to the reverse process of the opening operation, connecting the narrow gaps to make it an integr...

technical field

[0001] The invention relates to a layout analysis technology for pictures existing in image form in a digital library, in particular to a layout analysis technology for region division and classification. Background technique

[0002] With the research and development of computer and network technology, digital libraries are gradually developing from information-based processing and simple human-machine interface to knowledge-based processing and extensive understanding between machines, so that people can use computers and networks The ability to expand intellectual activities on a larger scale plays an extremely important role in all areas that need to communicate, disseminate, store, and utilize knowledge, including e-commerce, education, and telemedicine.

[0003] Since the books in the CADAL digital library exist in the form of images, they need to be processed by OCR to recognize the text and analyze the layout before they can provide in-depth services....

Examples

Embodiment

[0199] Below in conjunction with the method of the present invention, describe in detail the concrete steps that this example implements, take a certain page in the electronic book that CADAL library digital library scans as example here, as Figure 5 show, explain image 3 all the processes.

[0200] 1) Read the original image, perform grayscale processing on the color image, and read in the grayscale image

[0201] 2) Extract the edge of the picture. Create an 8*8 square structuring element to dilate the image. The image background is estimated by using the morphological opening operation imopen and closing operation imclose. The opening operation is to smooth the image contour, disconnect the narrow connection so as to realize the deletion of small protrusions, and construct a disc structure with a radius of 5 during the opening operation. The closing operation is equivalent to the reverse process of the opening operation, connecting the narrow gaps to make it an integr...

Images