Author: Ma Jian
E-mail: [email protected]
Published: 2020.01.04
These days in gxsd and eshuyuan have encountered some people talk about scanning e-book deal, a lot of people's habits is to use a general-purpose image processing software, including Photoshop, light and shadow magic hand, Mito Xiu Xiu, etc. to process the scanned version of the e-book. But in my opinion, scanning e-books and everyday landscapes, portraits entirely two things, general-purpose image processing software although powerful, but in the process the scanned e-book Shique will face powerful nowhere to make, want function and no dilemma, can not be compared with the scanning software designed to handle e-book developed. Of course, if the process of scanning e-book software to deal with daily photos, mostly belong head is filled.
He said rumor, to use when dealing with a few issues frequently encountered scan e-book, software, test it fineness of it:
1, the automatic correction feature oblique
scanned image is skewed is very common, right? This feature is important is not it? PS and other manual Straightening, automatic Straightening have it? ScanTailor (ST), ComocEnhancer Pro ( CEP) has.
2, automatic version of the heart center
Without the cut books for scanning, sweep out the contents of the version of the heart left, center-right is an inevitable thing, but in the general-purpose image editing software, the version of the heart center requires multiple steps: first cutting, then expansion side. When trimming seams encountered in the shadows, black border, interference pits, etc., automatically trimming numb claws, and can only rely on people. This software can scan e-book deal specifically is concerned, belong to the basic skills necessary okay?
3, automatic correction of uneven illumination of an image
for a flatbed scanner, the problem is generally not serious, the seam portion only will be some thick books, nausea burst will hit. For using a mobile phone or camera can shoot books, uniform illumination, no correction I've seen only in the case of the V-shaped scanner dozens of sets of their own do not have a shot, because the environment is not substantially daily photo there may be so uniform, intense lighting. Without correcting uneven illumination, the image is transferred directly to the binarized image is captured substantially dream or nightmare. PS can manually uneven illumination correction, automatic correction I have not seen live for a long time, but for ST, CEP this is a thing? ST is a fully automatic, CEP mouse to choose what parameters only.
4, the local adaptive binarization
in "The Scan and Share tutorial", the books should be repeatedly stressed with 300 DPI scan, then treated with software to 600 DPI. But this article is, after all, a few years ago, and considering the recent and future trends display resolution, I now handle plain text pages are basically 800 DPI base, even to the high 1200 DPI. Enlarged from 300 DPI to more than 800 DPI, inevitably there will be blurred, and so this time then the global Otsu binarization algorithm, such as control of the thickness of strokes will be powerless. General image processing software only global binarization, no local adaptive binarization. CEP in this regard basically alone this one.
5, the high-magnification high-order sharpening
after high magnification, the Gaussian blur + local adaptive binarization may control the stroke thickness and smoothness, but nothing blocking the stroke, sharpening rely on higher order hard knock. PS sharpening is too gentle, and CEP Gaussian sharpening immeasurably.
6, automatic deburring, automatically to pitting
Gaussian blur can be suitably smooth stroke prior to the binarized image, binarization, isolated point is removed, but some of the burrs and color binarization process resulting in deeper stroke caused by pitting is indelible. If GM PS treatment, pitting only with the mouse by removing burrs on the stroke only amplified by-pixel removal, which people have the patience? Automatic deburring, automatically to pitting of the CEP, all in terms of standard ST, ST is fully automatic but, can not perform detailed control, the CEP may need to manually configure the parameters. Examples of deburring can be seen here: https://www.cnblogs.com/stronghorse/p/7224976.html
7, to automatically short, the wavy line
are more classical the stuff inside, below the place names (horizontal) or left (vertical) have, little less modern text, but the header, footer and the body sometimes there will be those naked eye all right, but would interfere with OCR, so if you want to OCR, the thing is to remove the job. By manually sliding the mouse on the PS, Del key press bar, the CEP can be automatically removed in accordance with the parameters for the binarized image.
8, two-page automatic segmentation
plate when the scan order to improve efficiency, even the two-page sweep operation routine, but when reading or watching desired page by page, the paging processing for scanning books also are routine operations. In the case where the seam location is not fixed, PS paging only by hand, ST can be cut automatically by the algorithm.
9, surface flattening
in the library holding a cell phone filmed shared equally pages on the desktop people are very eager for this function, PS only provides keystone correction, but no page curvature correction, ST has.