#include <string.h>
#include "allheaders.h"

Macros
#define	L_BUF_SIZE 512

Functions
static l_int32	testLineAlignmentX (NUMA na1, NUMA na2, l_int32 shiftx, l_int32 delx, l_int32 nperline)

static l_int32	countAlignedMatches (NUMA nai1, NUMA nai2, NUMA nasx, NUMA nasy, l_int32 n1, l_int32 n2, l_int32 delx, l_int32 dely, l_int32 nreq, l_int32 *psame, l_int32 debugflag)

static void	printRowIndices (l_int32 index1, l_int32 n1, l_int32 index2, l_int32 n2)

l_ok	jbCorrelation (const char dirin, l_float32 thresh, l_float32 weight, l_int32 components, const char rootname, l_int32 firstpage, l_int32 npages, l_int32 renderflag)

l_ok	jbRankHaus (const char dirin, l_int32 size, l_float32 rank, l_int32 components, const char rootname, l_int32 firstpage, l_int32 npages, l_int32 renderflag)

JBCLASSER *	jbWordsInTextlines (const char dirin, l_int32 reduction, l_int32 maxwidth, l_int32 maxheight, l_float32 thresh, l_float32 weight, NUMA *pnatl, l_int32 firstpage, l_int32 npages)

l_ok	pixGetWordsInTextlines (PIX pixs, l_int32 minwidth, l_int32 minheight, l_int32 maxwidth, l_int32 maxheight, BOXA pboxad, PIXA ppixad, NUMA *pnai)

l_ok	pixGetWordBoxesInTextlines (PIX pixs, l_int32 minwidth, l_int32 minheight, l_int32 maxwidth, l_int32 maxheight, BOXA pboxad, NUMA *pnai)

l_ok	pixFindWordAndCharacterBoxes (PIX pixs, BOX boxs, l_int32 thresh, BOXA pboxaw, BOXAA pboxaac, const char *debugdir)

NUMAA *	boxaExtractSortedPattern (BOXA boxa, NUMA na)

l_ok	numaaCompareImagesByBoxes (NUMAA naa1, NUMAA naa2, l_int32 nperline, l_int32 nreq, l_int32 maxshiftx, l_int32 maxshifty, l_int32 delx, l_int32 dely, l_int32 *psame, l_int32 debugflag)

Variables
static const l_int32	JB_WORDS_MIN_WIDTH = 5

static const l_int32	JB_WORDS_MIN_HEIGHT = 3

Detailed Description


     Top-level jb2 correlation and rank-hausdorff
        l_int32         jbCorrelation()
        l_int32         jbRankHaus()

     Extract and classify words in textline order
        JBCLASSER      *jbWordsInTextlines()
        l_int32         pixGetWordsInTextlines()
        l_int32         pixGetWordBoxesInTextlines()

     Extract word and character bounding boxes
        l_int32         pixFindWordAndCharacterBoxes()

     Use word bounding boxes to compare page images
        NUMAA          *boxaExtractSortedPattern()
        l_int32         numaaCompareImagesByBoxes()
        static l_int32  testLineAlignmentX()
        static l_int32  countAlignedMatches()
        static void     printRowIndices()

Definition in file classapp.c.

Macro Definition Documentation

◆ L_BUF_SIZE

#define L_BUF_SIZE 512

size of filename buffer

Definition at line 59 of file classapp.c.

Referenced by fhmtautogen1(), fhmtautogen2(), fmorphautogen1(), jbCorrelation(), jbRankHaus(), makeBarrelshiftString(), makeBarrelshiftString(), sarrayMakeInnerLoopDWACode(), sarrayMakeInnerLoopDWACode(), selaAddBasic(), selaAddCrossJunctions(), selaAddDwaCombs(), selaAddDwaLinear(), selaAddTJunctions(), zlibCompress(), and zlibUncompress().

Function Documentation

◆ boxaExtractSortedPattern()

NUMAA * boxaExtractSortedPattern	(	BOXA *	boxa,
		NUMA *	na )

boxaExtractSortedPattern()

Parameters

[in]	boxa	typ. of word bounding boxes, in textline order
[in]	na	index of textline for each box in boxa

Returns: naa NUMAA, where each numa represents one textline, or NULL on error

Notes:
     (1) The input is expected to come from pixGetWordBoxesInTextlines().
     (2) Each numa in the output consists of an average y coordinate
         of the first box in the textline, followed by pairs of
         x coordinates representing the left and right edges of each
         of the boxes in the textline.

Definition at line 658 of file classapp.c.

References L_CLONE, and L_INSERT.

◆ countAlignedMatches()

static l_ok countAlignedMatches	(	NUMA *	nai1,
		NUMA *	nai2,
		NUMA *	nasx,
		NUMA *	nasy,
		l_int32	n1,
		l_int32	n2,
		l_int32	delx,
		l_int32	dely,
		l_int32	nreq,
		l_int32 *	psame,
		l_int32	debugflag )

static

Definition at line 927 of file classapp.c.

◆ jbCorrelation()

l_ok jbCorrelation	(	const char *	dirin,
		l_float32	thresh,
		l_float32	weight,
		l_int32	components,
		const char *	rootname,
		l_int32	firstpage,
		l_int32	npages,
		l_int32	renderflag )

jbCorrelation()

Parameters

[in]	dirin	directory of input images
[in]	thresh	typically ~0.8
[in]	weight	typically ~0.6
[in]	components	JB_CONN_COMPS, JB_CHARACTERS, JB_WORDS
[in]	rootname	for output files
[in]	firstpage	0-based
[in]	npages	use 0 for all pages in dirin
[in]	renderflag	1 to render from templates; 0 to skip

Returns: 0 if OK, 1 on error

Notes:
     (1) The images must be 1 bpp.  If they are not, you can convert
         them using convertFilesTo1bpp().
     (2) See prog/jbcorrelation for generating more output (e.g.,
         for debugging)

Definition at line 99 of file classapp.c.

References L_BUF_SIZE, and L_CLONE.

◆ jbRankHaus()

l_ok jbRankHaus	(	const char *	dirin,
		l_int32	size,
		l_float32	rank,
		l_int32	components,
		const char *	rootname,
		l_int32	firstpage,
		l_int32	npages,
		l_int32	renderflag )

jbRankHaus()

Parameters

[in]	dirin	directory of input images
[in]	size	of Sel used for dilation; typ. 2
[in]	rank	rank value of match; typ. 0.97
[in]	components	JB_CONN_COMPS, JB_CHARACTERS, JB_WORDS
[in]	rootname	for output files
[in]	firstpage	0-based
[in]	npages	use 0 for all pages in dirin
[in]	renderflag	1 to render from templates; 0 to skip

Returns: 0 if OK, 1 on error

Notes:
     (1) See prog/jbrankhaus for generating more output (e.g.,
         for debugging)

Definition at line 179 of file classapp.c.

References L_BUF_SIZE, and L_CLONE.

◆ jbWordsInTextlines()

JBCLASSER * jbWordsInTextlines	(	const char *	dirin,
		l_int32	reduction,
		l_int32	maxwidth,
		l_int32	maxheight,
		l_float32	thresh,
		l_float32	weight,
		NUMA **	pnatl,
		l_int32	firstpage,
		l_int32	npages )

jbWordsInTextlines()

Parameters

[in]	dirin	directory of input pages
[in]	reduction	1 for full res; 2 for half-res
[in]	maxwidth	of word mask components, to be kept
[in]	maxheight	of word mask components, to be kept
[in]	thresh	on correlation; 0.80 is reasonable
[in]	weight	for handling thick text; 0.6 is reasonable
[out]	pnatl	numa with textline index for each component
[in]	firstpage	0-based
[in]	npages	use 0 for all pages in dirin

Returns: classer for the set of pages

Notes:
     (1) This is a high-level function.  See prog/jbwords for example
         of usage.
     (2) Typically, use input of 75 - 150 ppi for finding words.

Definition at line 265 of file classapp.c.

References JbClasser::h, JB_WORDS_MIN_HEIGHT, JB_WORDS_MIN_WIDTH, L_NOCOPY, pixGetWordsInTextlines(), JbClasser::safiles, and JbClasser::w.

◆ numaaCompareImagesByBoxes()

l_ok numaaCompareImagesByBoxes	(	NUMAA *	naa1,
		NUMAA *	naa2,
		l_int32	nperline,
		l_int32	nreq,
		l_int32	maxshiftx,
		l_int32	maxshifty,
		l_int32	delx,
		l_int32	dely,
		l_int32 *	psame,
		l_int32	debugflag )

numaaCompareImagesByBoxes()

Parameters

[in]	naa1	for image 1, formatted by boxaExtractSortedPattern()
[in]	naa2	for image 2, formatted by boxaExtractSortedPattern()
[in]	nperline	number of box regions to be used in each textline
[in]	nreq	number of complete row matches required
[in]	maxshiftx	max allowed x shift between two patterns, in pixels
[in]	maxshifty	max allowed y shift between two patterns, in pixels
[in]	delx	max allowed difference in x data, after alignment
[in]	dely	max allowed difference in y data, after alignment
[out]	psame	1 if nreq row matches are found; 0 otherwise
[in]	debugflag	1 for debug output

Returns: 0 if OK, 1 on error

Notes:
     (1) Each input numaa describes a set of sorted bounding boxes
         (sorted by textline and, within each textline, from
         left to right) in the images from which they are derived.
         See boxaExtractSortedPattern() for a description of the data
         format in each of the input numaa.
     (2) This function does an alignment between the input
         descriptions of bounding boxes for two images. The
         input parameter nperline specifies the number of boxes
         to consider in each line when testing for a match, and
         nreq is the required number of lines that must be well-aligned
         to get a match.
     (3) Testing by alignment has 3 steps:
         (a) Generating the location of word bounding boxes from the
             images (prior to calling this function).
         (b) Listing all possible pairs of aligned rows, based on
             tolerances in horizontal and vertical positions of
             the boxes.  Specifically, all pairs of rows are enumerated
             whose first nperline boxes can be brought into close
             alignment, based on the delx parameter for boxes in the
             line and within the overall the maxshiftx and maxshifty
             constraints.
         (c) Each pair, starting with the first, is used to search
             for a set of nreq - 1 other pairs that can all be aligned
             with a difference in global translation of not more
             than (delx, dely).

Definition at line 744 of file classapp.c.

References L_CLONE.

◆ pixFindWordAndCharacterBoxes()

l_ok pixFindWordAndCharacterBoxes	(	PIX *	pixs,
		BOX *	boxs,
		l_int32	thresh,
		BOXA **	pboxaw,
		BOXAA **	pboxaac,
		const char *	debugdir )

pixFindWordAndCharacterBoxes()

Parameters

[in]	pixs	2, 4, 8 or 32 bpp; colormap OK; typ. 300 ppi
[in]	boxs	[optional] region to select in pixs
[in]	thresh	binarization threshold (typ. 100 - 150)
[out]	pboxaw	return the word boxes
[out]	pboxaac	return the character boxes
[in]	debugdir	[optional] for debug images; use NULL to skip

Returns: 0 if OK, 1 on error

Notes:
     (1) If boxs == NULL, the entire input image is used.
     (2) Having an input pix that is not 1bpp is necessary to reduce
         touching characters by using a low binarization threshold.
         Suggested thresholds are between 100 and 150.
     (3) The coordinates in the output boxes are global, with respect
         to the input image.

Definition at line 516 of file classapp.c.

References L_COPY, L_INSERT, L_SELECT_IF_BOTH, L_SELECT_IF_GTE, L_SORT_BY_X, L_SORT_INCREASING, and pixGetWordBoxesInTextlines().

◆ pixGetWordBoxesInTextlines()

l_ok pixGetWordBoxesInTextlines	(	PIX *	pixs,
		l_int32	minwidth,
		l_int32	minheight,
		l_int32	maxwidth,
		l_int32	maxheight,
		BOXA **	pboxad,
		NUMA **	pnai )

pixGetWordBoxesInTextlines()

Parameters

[in]	pixs	1 bpp, typ. 75 - 150 ppi
[in]	minwidth	of saved components; smaller are discarded
[in]	minheight	of saved components; smaller are discarded
[in]	maxwidth	of saved components; larger are discarded
[in]	maxheight	of saved components; larger are discarded
[out]	pboxad	word boxes sorted in textline line order
[out]	pnai	[optional] index of textline for each word

Returns: 0 if OK, 1 on error

Notes:
     (1) The input should be at a resolution of between 75 and 150 ppi.
     (2) This is a special version of pixGetWordsInTextlines(), that
         just finds the word boxes in line order, with a numa
         giving the textline index for each word.
         See pixGetWordsInTextlines() for more details.

Definition at line 452 of file classapp.c.

References L_CLONE.

Referenced by pixFindWordAndCharacterBoxes().

◆ pixGetWordsInTextlines()

l_ok pixGetWordsInTextlines	(	PIX *	pixs,
		l_int32	minwidth,
		l_int32	minheight,
		l_int32	maxwidth,
		l_int32	maxheight,
		BOXA **	pboxad,
		PIXA **	ppixad,
		NUMA **	pnai )

pixGetWordsInTextlines()

Parameters

[in]	pixs	1 bpp, typ. 75 - 150 ppi
[in]	minwidth	of saved components; smaller are discarded
[in]	minheight	of saved components; smaller are discarded
[in]	maxwidth	of saved components; larger are discarded
[in]	maxheight	of saved components; larger are discarded
[out]	pboxad	word boxes sorted in textline line order
[out]	ppixad	word images sorted in textline line order
[out]	pnai	index of textline for each word

Returns: 0 if OK, 1 on error

Notes:
     (1) The input should be at a resolution of between 75 and 150 ppi.
     (2) The four size constraints on saved components are all
         scaled by reduction.
     (3) The result are word images (and their b.b.), extracted in
         textline order, at either full res or 2x reduction,
         and with a numa giving the textline index for each word.
     (4) The pixa and boxa interfaces should make this type of
         application simple to put together.  The steps are:
          ~ generate first estimate of word masks
          ~ get b.b. of these, and remove the small and big ones
          ~ extract pixa of the word images, using the b.b.
          ~ sort actual word images in textline order (2d)
          ~ flatten them to a pixa (1d), saving the textline index
            for each pix
     (5) In an actual application, it may be desirable to pre-filter
         the input image to remove large components, to extract
         single columns of text, and to deskew them.  For example,
         to remove both large components and small noisy components
         that can interfere with the statistics used to estimate
         parameters for segmenting by words, but still retain text lines,
         the following image preprocessing can be done:
               Pix *pixt = pixMorphSequence(pixs, "c40.1", 0);
               Pix *pixf = pixSelectBySize(pixt, 0, 60, 8,
                                    L_SELECT_HEIGHT, L_SELECT_IF_LT, NULL);
               pixAnd(pixf, pixf, pixs);  // the filtered image
         The closing turns text lines into long blobs, but does not
         significantly increase their height.  But if there are many
         small connected components in a dense texture, this is likely
         to generate tall components that will be eliminated in pixf.

Definition at line 377 of file classapp.c.

References L_CLONE, and L_COPY.

Referenced by jbWordsInTextlines().

◆ printRowIndices()

static void printRowIndices	(	l_int32 *	index1,
		l_int32	n1,
		l_int32 *	index2,
		l_int32	n2 )

static

Definition at line 1014 of file classapp.c.

◆ testLineAlignmentX()

static l_int32 testLineAlignmentX	(	NUMA *	na1,
		NUMA *	na2,
		l_int32	shiftx,
		l_int32	delx,
		l_int32	nperline )

static

Definition at line 877 of file classapp.c.

Variable Documentation

◆ JB_WORDS_MIN_HEIGHT

const l_int32 JB_WORDS_MIN_HEIGHT = 3

static

min. word height in pixels

Definition at line 61 of file classapp.c.

Referenced by jbWordsInTextlines().

◆ JB_WORDS_MIN_WIDTH

const l_int32 JB_WORDS_MIN_WIDTH = 5

static

min. word width in pixels

Definition at line 60 of file classapp.c.

Referenced by jbWordsInTextlines().

Macros

Functions

Variables

Detailed Description

Macro Definition Documentation

◆ L_BUF_SIZE

Function Documentation

◆ boxaExtractSortedPattern()

◆ countAlignedMatches()

◆ jbCorrelation()

◆ jbRankHaus()

◆ jbWordsInTextlines()

◆ numaaCompareImagesByBoxes()

◆ pixFindWordAndCharacterBoxes()

◆ pixGetWordBoxesInTextlines()

◆ pixGetWordsInTextlines()

◆ printRowIndices()

◆ testLineAlignmentX()

Variable Documentation

◆ JB_WORDS_MIN_HEIGHT

◆ JB_WORDS_MIN_WIDTH