TWI396099B - The typographical method of video text file - Google Patents

Description

Typesetting method for video text files

The present invention is a typesetting method for a typeface of an image text file, in particular, a typesetting method for a typeface image file that can be automatically typeset and lined up according to the size of the reading interface of different vehicles.

E-books or electronic books refer to e-readers and a collective term for their digital content. They are the trend of information loaders in recent years and gradually replace traditional paper books. According to the definition of the Oxford Dictionary, the so-called "e-book" refers to "a printed book of electronic versions that can be read on a personal computer or a hand-held instrument." The above-mentioned hardware reading interface is generally called an "e-book reader", and a personal computer and some mobile phones can also be used as an e-book reader. Because e-books can be read by different types of e-book readers (or formerly known as vehicles), digital files are produced in digital publishing, and reading in different e-book readers becomes an important key. .

In the past, book management, because at the beginning of the development of the computer, most of the old books still use the ink printing method of the ink, there is no corresponding editable electronic file format to provide loading and reading. For these old books, the current general method is to scan the book into an image file and then load it. However, if the scanned files are directly loaded into the e-reader and cannot be automatically changed according to the size adjustment of the reading interface, it is necessary to frequently use the scrolling function of the up and down or left and right scrolls or change the viewing size function. The purpose of reading the entire page is that it is inconvenient and time consuming to read.

In order to solve these problems, the conventional technology converts the scanned files into an editable electronic file format, and creates a digital file for editing, and then loads and reads. The electronic translation technology is combined with optical character recognition (OCR), and the text image captured by the scanner is electronically translated into an editable file, and the editable file can be read according to the size of the reading interface. And adjust the line break.

A preferred optical character recognition system must have human-like reading and good recognition of different printed fonts. Generally speaking, in English identification, due to the limited number of English letters and text symbols, the optical character recognition system has better recognition effect on the text files generated after typesetting, and can be expected to reach nearly 100%. However, for ideographic languages or block languages, such as traditional Chinese, Japanese, Korean, etc., the electronic translation performance of the optical character recognition system is unsatisfactory due to the large text database and complex font style. To achieve a readable correct rate, it must be proofed manually. In the case of a large number of old books, it takes a considerable amount of time and expense to scan and electronically translate each book into an editable file and then correct the results of the electronic translation. Moreover, if the demand is only for reading, and the file editing is not required, the electronic identification does not meet the existing needs.

In summary, the present invention provides a movable typesetting method for an image text file, and an object of the present invention is to provide a movable typesetting of an image text file that can be automatically typeset and lined according to the reading interface size of different carriers.

The present invention provides a movable typesetting method for video text files, and another object thereof is to improve display speed and convenience; the movable typesetting method of the image text file of the present invention does not need to be compared with individual character pairs of the optical character recognition system, but directly The image text file is properly cut and quickly displayed on different sizes of the reading interface, so there is no need for a large amount of proofing workmanship after the optical character recognition system is processed, and there is no inconvenience and lack of flexibility in direct presentation of the image text file.

The movable typesetting method for the image text file provided by the present invention comprises the following steps: according to the reading order, Performing a first direction of cutting on an image text file, causing the text file to form a plurality of text string images; performing a second direction cutting on each of the text string images, such that the image text file forms a plurality of independent character characters Image; based on a reading interface, the independent character images are typeset, so that the layout layout can automatically wrap according to the size of the reading interface.

In one embodiment, the present invention records a black pixel array and a black pixel array of the image text file, thereby finding the upper, lower, left and right borders of each character, and thus each character in the image text file. Split and redetype in characters, achieving the ability to automatically wrap depending on the size of the reading interface.

In one embodiment, the present invention stores the reading order and segmentation information in the original image text file and redisplays it in the layout program.

w‧‧‧Reading interface and display buffer width

h‧‧‧Reading interface height

H‧‧‧Display buffer height

y‧‧‧The height has been shown last time

X‧‧‧X coordinate axis

Y‧‧‧Y coordinate axis

1~5, 11~13, 21, 24~25, 51~57‧‧‧ steps

221~224, 231~234, 521~527, 531~535‧‧‧ steps

The above and other objects, features, advantages and embodiments of the present invention will become more readily understood. The flow chart of the embodiment; the second figure shows a flow chart of the embodiment of the invention for cutting the first direction of the image text file to form a plurality of character strings; the third figure shows that the image text file is cut into independent characters after the invention a flow chart of an embodiment of a movable type layout in a reading interface; a fourth figure showing a flow chart of one of the sub-steps included in the step of "writing the display buffer" of the present invention; and the fifth figure shows the printing of the present invention. A flow chart of one of the sub-steps included in the step of displaying a buffer; a sixth A picture showing the display buffer range when the buffer data is in the end; and a sixth B picture showing the display buffer when the buffer data is not at the end range.

The following description and examples are presented to explain the details of the invention. However, those skilled in the art should readily appreciate that the described embodiments are susceptible to variations and modifications. Therefore, the embodiments of the afterload are not intended to limit the scope of protection of the present invention. The movable typesetting method of the image text file of the present invention substantially includes two main programs of "text cutting" and "live typesetting". In the first "text cutting" program, the present invention cuts the image text file into a plurality of independent characters and stores each of the characters in an image format and related information. In the following "Live Typesetting" program, the present invention performs typesetting on the cut characters in units of characters, so that the reading interface of different sizes can be typeset and automatically changed to achieve fast and convenient. Reader of the reading.

The "image text file" described in the present invention is defined as a file format that cannot be directly edited, for example, a pdf file or a tiff file, but is not limited to the examples. In addition, the name of the present invention is based on the image text file, but the practice is not limited to the fact that the overall file format is a text. Please refer to the first figure, which shows a flow chart of an embodiment of "text cutting" in the movable typesetting method of the image text file of the present invention, which comprises: Step 1: "Get an image text file", Step 2: "Does The image text file is cut in a first direction to form a plurality of text string images. Step 3: "Cutting a second direction for each of the text strings to form a plurality of characters", step 4: Each character is stored as an image, and the reading order and segmentation information are retained" and step 5: "Live typesetting".

First, in the movable typesetting method of the image text file of the present invention, an image text file must be obtained (step 1), and when the image text file is obtained, the width W1 and the height H1 of the image are simultaneously captured. Then, according to the order of reading, the image text file is cut in a first direction to form a plurality of text string images (step 2), and then one of each character string is successively performed according to the order of the text string reading. Cutting in the second direction to form a plurality of characters (step 3). After cutting, each character can be stored as an image, and the original reading order and segmentation information are retained (step 4), and The character is a unit, and the typesetting is performed on the individual characters after cutting (step 5).

In an embodiment, in order to reduce the storage capacity and the processing time, the step of adding an image file to the step 1: "transforming the image file into a black and white binary image" is added to the step 1: The size of the file is only processed for black and white binary files. The "black and white binary image" refers to all the colors in the image in black or white, so it is called a black and white binary image.

Furthermore, in order to simplify the cutting process, the cutting order of the straight or horizontal files is the same, that is, the horizontal cutting is performed first, and then the straight cutting is performed; before the cutting of the step 2, the step 12 may be included: The program of the typesetting direction, and based on the confirmed text layout direction, performs a step 13 of "rotating an appropriate angle of the image text file". The "rotation of an appropriate angle" is exemplified. If the original image text file is straight and the right-to-left text string is read, the "rotation of the appropriate angle" may be rotated counterclockwise. Ten degrees. Therefore, after performing the appropriate rotation angle, regardless of whether the original text layout direction is straight or horizontal, the horizontal direction can be unified as the "first direction" and the straight direction as the "second direction". Simplification of the cutting program.

However, in other embodiments, the "first direction" and the "second direction" described in steps 2 and 3 are freely defined according to the layout direction and are not limited. For example, if the text layout of the original image text file is horizontal, the first direction may be defined as a horizontal direction, and the second direction may be a straight direction. Conversely, if the text layout of the original image text file is straight, the first direction can be defined as straight direction, and the second direction can be horizontal. Therefore, regardless of the straight layout or the horizontal layout, after the "first direction" and the "second direction" are cut, an image text file can form a plurality of independent character images.

Please refer to the second figure, which shows a flow chart of an embodiment of the present invention for performing a first direction cut on an image text file to form a plurality of text strings. In this embodiment, the present invention scans and reads the black portion of the image, thereby defining the upper and lower boundaries of each character string, and dividing the image by the upper and lower boundaries. The text file becomes a plurality of text strings; however, in practice, substantially equal division can be performed using other methods, such as reading a white portion of the image for segmentation, and the present invention is not limited by the segmentation of the disclosed embodiment. Furthermore, for convenience of description, in the second figure, it is assumed that the first direction cut on the image text file is a horizontal cut, and the black and white binary map conversion in step 11 is performed.

Referring to the second figure, first, as in step 21: initializing, and storing the number of black pixels in each pixel column of the image into an array, represented as a black pixel array BR(i), where i is The index number of the prime column, the order of increasing the index number is determined according to the reading order. For example, BR(8)=1000 represents that the number of black pixels stored in the pixel column of i=8 is 1000. By recording the number of black pixels in the image, you can know the location of the characters in the image file and its boundaries. In an embodiment of the present invention, considering noises such as stains on the old books, the pre-processing of the files may be performed before the recording of the black pixels. Decrease, so that it does not affect the reading of the correct character.

Continually, in order to implement the cut image file as a plurality of character columns, it is necessary to find the upper and lower boundaries of each character string. The BR(i) value of each pixel column in the image is sequentially considered (step 221), if the BR(i) value of one of the pixel columns is zero (that is, the BR(i) array is blank, below This is called a zero element), and the value of its next pixel column, BR(i+1), is greater than zero (step 222), which means that the pixel number of the index number (i+1) is all white. The first black pixel column immediately after the prime column, so the pixel column can be regarded as the starting boundary of the j-th character string, and the information of the pixel column is recorded as the upper boundary U(j (Step 224); where j is the index number of a text string. However, if the condition of step 222 is not satisfied, the confirmation of the next pixel column is continued (step 223). In addition, if an image text file has been converted by the black and white binary image of step 11 and subjected to pre-de-noise processing, in step 222, it may be considered to find the element of the BR matrix that is greater than zero after the zero element, thereby simplifying to find out The BR matrix is a non-zero element after the zero element.

After finding the boundary above the j-th character string, the lower boundary of the text string can be found in a similar manner: the BR(i) value of each pixel column in the image after the upper boundary is found is sequentially considered (step 231). As its The BR(i) value of the first pixel column is greater than zero, and the value BR(i+1) of the next pixel column is exactly zero (step 232), indicating that the index number (i+1) pixel is listed as The black pixel sequence is followed by the first all-white color column (zero element), so the pixel column can be regarded as the end boundary of the j-th character string, and the information of the pixel column is recorded as The boundary L(j) (step 234). However, if the condition of step 232 is not satisfied, the confirmation of the next pixel column is continued (step 233). Then, all the pixel columns are considered until the end of the file (step 24), and the confirmation of the upper and lower boundaries of each character string is ended (step 25). Similar to the procedure for finding the upper boundary, for example, an image text file has been converted by the black and white binary image of step 11 and subjected to pre-de-noise processing. In step 232, it may be considered to find the BR matrix after the greater than zero element. The zero element is simplified to find the zero element of the BR matrix after the non-zero element.

In order to cope with noise problems such as stains in the image file, the above conditions may be relaxed to set a threshold k, which is regarded as BR(i)<k when determining the lower boundary above the pixel column. An all-white color without text, without having to be a zero element (ie BR(i)=0). For the cutting in the second direction, a cutting similar to the first direction in the second drawing can be performed: for each text string image, the black pixels of each column of the text string are sequentially recorded to an array. , denoted as a black pixel array BC(m), where m is the column index number of a text string. Consider m in the order of reading. Suppose the character reading order of a character string is from left to right. When m satisfies BC(m)=0 and BC(m+1)>0, the left boundary of a character can be found. Considering the image on the right side of the character n>m, where n is also the column index number of the character string. If n satisfies BC(n)>0 and BC(n+1)=0, the character can be found. Right boundary R. In conjunction with the upper and lower boundaries of the preceding character string, a separate character can be defined and cut.

Considering some special character cases, the present invention can add some post processing to correct the error cut and make the output result correct. For example, in the case of outputting some merged block characters, such as "North", a portion of the central portion of the "North" character may cause a segmentation error. Therefore, in one embodiment of the present invention, a merge coefficient may be set to perform the The combined adjustment of some special characters. For example, after the character is cut, if the left and right boundaries of a character are L, R, and the upper and lower boundaries are U, L, and (R-L) < a × (L - U), that is, a ratio of the left and right widths of the divided characters to the upper and lower heights is small; wherein a is a combining coefficient, for example, 3/4. And the next character of the character, the left and right boundaries are L', R' and the upper and lower boundaries are U', L', and (R'-L') < a × (L'-U'), that is, The left and right widths of the next character of the character are also smaller than the upper and lower heights. In other words, if the left and right widths of the two characters are smaller than the upper and lower heights, then the two characters before and after can be combined and output. In step 4 of the present invention, after the image is cut into independent characters, the character reading order and the original paragraph information must be considered and recorded, in order to re-read the character reading order and segmentation in the "live typesetting" of step 5. Information retention is presented. In general, the appearance of a new paragraph, for example, may be related to: (a) a text string is more spaced from the previous text string; (b) the beginning of a text string includes a blank indentation; Or (c) the end of a text string includes a blank, etc., but is not limited to the examples.

In the case of (a), the present invention can determine whether the difference between the upper boundary U(i+1) of the next character string and the lower boundary L(i) of the previous character string is greater than a threshold t1, indicating It is U(i+1)-L(i)>t1, where t1 can be set to a certain value, for example, the height of the text string image with the highest frequency of occurrence. In other words, if the distance between a text string and the previous text string is greater than the height of the text string with the highest frequency of occurrence, the two text strings are considered to belong to two paragraphs, and a paragraph symbol is outputted in the two text strings.

Alternatively, the segmentation data may be output in the case of (b): the first occurrence of the character is found in the i-th character string, that is, the black pixel first appears to make BC(m)>0, set. A blank limit t2, such as m>t2×(L(i)-U(i)), that is, the beginning of the character string is greater than t2 times the height of the text string image, then a paragraph symbol may be considered for output.

Alternatively, the segmentation data may be output in the case of (c): first find the end of the i-th text string, that is, BC(n)>0 and BC(n+1)=0, and define a final Limited to t3. For example, the image width W1 (obtained in step 1 of the first figure) is the portion of the character string, that is, the portion of the end of the text string, if the portion is greater than t3 times the height of the text string image, it is represented as W1- m>t3× ((L(i)-U(i)), you can consider outputting a paragraph symbol.

So far, the present invention has cut an image text file into a plurality of independent characters, and stores the required reading order, boundary information, and paragraph information, etc., for use in the typesetting of step 5. The third figure shows a flow chart of an embodiment of the present invention in which the image text file is cut into independent characters and then re-formatted into a reading interface, which is a sub-step of step 5 in the first figure. First, in order for the present invention to automatically typeset lines in response to the reading interface size of different vehicles, the width w and height h of the reading interface must be obtained first (step 51), and the width w of a display buffer and Height H; wherein the width of the reading interface of the vehicle refers to the width of the handheld reading state, and the display buffer refers to the size of the display random access memory (RAM) whose scroll is enabled. Here, for convenience of explanation, it is assumed that the width of the reading interface of the carrier is the same as the width of the display buffer, and is also w. In addition, the reading interface size of the vehicle can also be obtained and updated immediately after the user changes its settings.

After obtaining the reading frame of the vehicle and the width and height of the display buffer (step 51), the image information to be displayed must be written into the display buffer (step 52, connected to the fourth picture), and then printed to the display buffer ( Step 53, connect to the fifth figure), and update the range of image display (step 54). Subsequently, it is confirmed whether all the materials have been displayed or received the termination instruction (step 55), and if so, the layout display program is ended (step 56); if not, it is confirmed whether the changed image display range exceeds the display buffer Zone (step 57). If the changed image display range has exceeded the display buffer, it means that the data stored in the display buffer is insufficient to be provided and displayed on a complete page of the reading interface, so it is necessary to return to step 52 to continue writing the data to the buffer. For display and form a loop until all data has been displayed or needs to be terminated (step 54), then stop (step 55).

Following the fourth figure, in step 52, the present invention sequentially writes the divided character image data into the display buffer. Please refer to the fourth figure, which shows a flow chart of one of the sub-steps included in step 52. After step 51, the character image data currently to be displayed is first obtained sequentially (step 521). , the character image data is represented by A. Starting from the start image of an original page, the character image data A is initially the first image data to be written. Here, in order to simplify the description, the fourth figure is exemplified by the horizontal typesetting, but is not actually limited thereto.

After the display character image A is obtained (step 521), it is confirmed whether the current write is to be segmented (step 522). If segmentation is required, the image must be segmented before displaying the image A (step 523). If there is no need to change the line, it is necessary to confirm whether the same line space of the buffer has enough width and then write to the character image data A (step 524), and if there is no space, it must be changed (step 523). On the other hand, if the width of the buffer is enough to write the character image data A, the non-segmented image data A is directly written next to the original image that has been written (step 525), and then returns to step 521 to Form a loop. After the confirmation in step 522, if the segmentation or the buffer width is insufficient and the line needs to be changed, after the line feed (step 523), it is confirmed whether the display area height is greater than the allowable height H of the buffer after writing the image A (step 526), wherein the display buffer height H is obtained in step 51. If the buffer height after writing image A is less than the allowable height H of the buffer, it means that the writing can continue, so the writing can be continued in step 525. Otherwise, the space representing the display buffer is insufficient to continue writing, the current write operation must be stopped, and the current character image A is recorded (step 527), so that the next write operation can be started from the character image A. beginning. After step 527, proceed to step 53: print the data of the buffer.

Please refer to the fifth figure, which shows a flow chart of one of the sub-steps included in step 53. After completing the writing of the display buffer (step 52), it is necessary to confirm whether the data height currently in the display buffer is the same height as the reading interface (step 531). If the height is the same, the data in the buffer range can be printed directly on the reading interface (step 532). Otherwise, it is necessary to further confirm whether the data of the buffer has been completed (step 533), and determine the printing range of the buffer according to the reading interface and the size relationship of the buffer. If the data of the buffer has reached the bottom, as shown in Figure 6A, assuming that the height range is displayed to the y position last time (as indicated by the dotted area in Figure 6A), the height of the buffer is H (eg The dotted area in Figure 6A is shown by the white area), while the reading interface The height is h (as shown in the white area in Figure 6 plus the slash area), the reading interface can display the height h greater than the current required display height (Hy), expressed as h>Hy, in other words, the data display Already, so the buffer range printed this time is the area defined by the four coordinate points (0, y), (w, y), (0, H) and (w, H) (step 534), The positions of the coordinate points are relative to the coordinate axes X and Y shown in Figure 6A.

If the data of the buffer has not yet reached the bottom, as shown in Figure B, if the height of the buffer is H (as in the dotted area of Figure B plus the white area and the slash area), the data in the buffer has not yet been In the end, that is, the buffer height H will be greater than the height y of the displayed area plus the height h of the reading interface, indicating that H>y+h, then the buffer range of the current printing is the reading interface range (such as the sixth B picture). The area indicated by the white area, that is, the area defined by the four coordinate points (0, y), (w, y), (0, y + h), and (w, y + h) (step 535). After the buffer display area is printed, the range of image display is updated (step 54).

In the present invention, the manner of updating the image display is not limited, and the scrolling of the upper and lower scrolls or the function of "next page (page down)" can be used, and according to the manner selected above. , update the display range of the image. Taking the reel as an example, under the scrolling scroll, it is assumed that each scrolling causes the reading interface to move down the d range. If the starting coordinate of the reading interface is (0, y), the reading interface is changed after the image is displayed. The start coordinates are updated to (0, y + d). The movable typesetting method of the image text file of the present invention includes two main programs of "text cutting" and "live typesetting". In the "text cutting" program, the present invention cuts the image text file into a plurality of independent characters and stores each of the characters in an image format. In the following "Live Typesetting" program, the character of the present invention is a unit for performing typesetting on the cut characters, so that the reading interface of different sizes can be typeset and automatically wrapped for quick and convenient. Reader of the reading.

For those skilled in the art to which the present invention pertains, the various modifications of the present invention can be derived without departing from the spirit and scope of the invention. More as well as other specific embodiments. Therefore, the present invention is not to be limited to the specific embodiments disclosed herein.

1~5‧‧‧ steps

11~13‧‧‧Steps

Claims (9)

A movable typesetting method for image text files, the steps comprising: (a) drawing a width and a height of an image text file, and performing a first direction cutting according to the reading order, so that the image text file forms a plurality of text strings. (b) performing a second direction of cutting on each of the text string images in the order in which the text string images are read, such that the image text files form a plurality of independent character images; (c1) obtaining a reading interface Width and height and a width and height that can be written to the display buffer; (c2) writing the independent character image to the display buffer in reading order; (c3) printing according to the width and height of the reading interface The display buffer is to the reading interface; (c4) updating the image display range of the reading interface, and formatting the independent character images so that the layout layout can automatically wrap according to the reading interface size. For example, the movable typesetting method of the image text file mentioned in the first paragraph of the patent application includes a step (a0) before the step (a): converting the image text file into a black and white binary image. The movable typesetting method of the image text file of claim 2, wherein the first direction is a horizontal direction and the second direction is a straight direction. The movable typesetting method of the image text file described in claim 3, wherein the step (a) comprises: (a1) storing the number of black pixels of each pixel column in the image text file to a black drawing. a matrix of the prime array; (a2) finding the upper and lower boundaries of each of the character strings by the boundary between the zero elements and the non-zero elements in the black pixel array, thereby performing the cutting in the first direction. The movable typesetting method of the image text file described in claim 3, wherein the step (b) The method includes: (b1) storing the number of black pixels of each pixel column in each character string to a black pixel column array; (b2) the boundary between the zero element and the non-zero element in the black pixel column array, Find the left and right boundaries of each character in the string to perform the cutting in the second direction. The movable typesetting method of the image text file described in claim 4, wherein between (a2) and (b), a step (a3) is included: one of the text strings of the text string image If the distance between the upper boundary of the image and the lower boundary of the previous text string image is greater than a certain limit, a piece of segment information is recorded between the two character strings. The movable typesetting method of the image text file of claim 1, wherein in the step (c2), the width of the image written to the buffer in one line plus the width of one of the characters is larger than the reading interface. For the width, the character image is written on the next line. The movable typesetting method of the image text file according to claim 1, wherein in the step (c2), if the height of the character image written to the buffer plus the height of the character is greater than the buffer height, Stop writing and proceed to step (c3). The movable typesetting method of the image text file of claim 1, wherein in the step (c4), the image display range is updated by scrolling the scroll up or down or touching the upper and lower pages.