/*!
* pixaSelectByWidthHeightRatio()
*
* Input: pixas
* thresh (threshold ratio of width/height)
* type (L_SELECT_IF_LT, L_SELECT_IF_GT,
* L_SELECT_IF_LTE, L_SELECT_IF_GTE)
* &changed (<optional return> 1 if changed; 0 if clone returned)
* Return: pixad, or null on error
*
* Notes:
* (1) Returns a pixa clone if no components are removed.
* (2) Uses pix and box clones in the new pixa.
* (3) This filters components based on the width-to-height ratio
* of each pix.
* (4) Use L_SELECT_IF_LT or L_SELECT_IF_LTE to save components
* with less than the threshold ratio, and
* L_SELECT_IF_GT or L_SELECT_IF_GTE to remove them.
*/
PIXA *
pixaSelectByWidthHeightRatio(PIXA *pixas,
l_float32 thresh,
l_int32 type,
l_int32 *pchanged)
{
NUMA *na, *nai;
PIXA *pixad;
PROCNAME("pixaSelectByWidthHeightRatio");
if (!pixas)
return (PIXA *)ERROR_PTR("pixas not defined", procName, NULL);
if (type != L_SELECT_IF_LT && type != L_SELECT_IF_GT &&
type != L_SELECT_IF_LTE && type != L_SELECT_IF_GTE)
return (PIXA *)ERROR_PTR("invalid type", procName, NULL);
/* Compute component ratios. */
na = pixaFindWidthHeightRatio(pixas);
/* Generate indicator array for elements to be saved. */
nai = numaMakeThresholdIndicator(na, thresh, type);
numaDestroy(&na);
/* Filter to get output */
pixad = pixaSelectWithIndicator(pixas, nai, pchanged);
numaDestroy(&nai);
return pixad;
}
int main(int argc,
char **argv)
{
l_int32 w, h, n, i, sum, sumi, empty;
BOX *box1, *box2, *box3, *box4;
BOXA *boxa, *boxat;
NUMA *na1, *na2, *na3, *na4, *na5;
NUMA *na2i, *na3i, *na4i, *nat, *naw, *nah;
PIX *pixs, *pixc, *pixt, *pixt2, *pixd, *pixcount;
PIXA *pixas, *pixad, *pixac;
pixDisplayWrite(NULL, -1);
/* Draw 4 filled boxes of different sizes */
pixs = pixCreate(200, 200, 1);
box1 = boxCreate(10, 10, 20, 30);
box2 = boxCreate(50, 10, 40, 20);
box3 = boxCreate(110, 10, 35, 5);
box4 = boxCreate(160, 10, 5, 15);
boxa = boxaCreate(4);
boxaAddBox(boxa, box1, L_INSERT);
boxaAddBox(boxa, box2, L_INSERT);
boxaAddBox(boxa, box3, L_INSERT);
boxaAddBox(boxa, box4, L_INSERT);
pixRenderBox(pixs, box1, 1, L_SET_PIXELS);
pixRenderBox(pixs, box2, 1, L_SET_PIXELS);
pixRenderBox(pixs, box3, 1, L_SET_PIXELS);
pixRenderBox(pixs, box4, 1, L_SET_PIXELS);
pixt = pixFillClosedBorders(pixs, 4);
pixDisplayWrite(pixt, 1);
pixt2 = pixCreateTemplate(pixs);
pixRenderHashBox(pixt2, box1, 6, 4, L_POS_SLOPE_LINE, 1, L_SET_PIXELS);
pixRenderHashBox(pixt2, box2, 7, 2, L_POS_SLOPE_LINE, 1, L_SET_PIXELS);
pixRenderHashBox(pixt2, box3, 4, 2, L_VERTICAL_LINE, 1, L_SET_PIXELS);
pixRenderHashBox(pixt2, box4, 3, 1, L_HORIZONTAL_LINE, 1, L_SET_PIXELS);
pixDisplayWrite(pixt2, 1);
/* Exercise the parameters */
pixd = pixSelectBySize(pixt, 0, 22, 8, L_SELECT_HEIGHT,
L_SELECT_IF_GT, NULL);
count_pieces(pixd, 1);
pixd = pixSelectBySize(pixt, 0, 30, 8, L_SELECT_HEIGHT,
L_SELECT_IF_LT, NULL);
count_pieces(pixd, 3);
pixd = pixSelectBySize(pixt, 0, 5, 8, L_SELECT_HEIGHT,
L_SELECT_IF_GT, NULL);
count_pieces(pixd, 3);
pixd = pixSelectBySize(pixt, 0, 6, 8, L_SELECT_HEIGHT,
L_SELECT_IF_LT, NULL);
count_pieces(pixd, 1);
pixd = pixSelectBySize(pixt, 20, 0, 8, L_SELECT_WIDTH,
L_SELECT_IF_GT, NULL);
count_pieces(pixd, 2);
pixd = pixSelectBySize(pixt, 31, 0, 8, L_SELECT_WIDTH,
L_SELECT_IF_LT, NULL);
count_pieces(pixd, 2);
pixd = pixSelectBySize(pixt, 21, 10, 8, L_SELECT_IF_EITHER,
L_SELECT_IF_LT, NULL);
count_pieces(pixd, 3);
pixd = pixSelectBySize(pixt, 20, 30, 8, L_SELECT_IF_EITHER,
L_SELECT_IF_GT, NULL);
count_pieces(pixd, 2);
pixd = pixSelectBySize(pixt, 22, 32, 8, L_SELECT_IF_BOTH,
L_SELECT_IF_LT, NULL);
count_pieces(pixd, 2);
pixd = pixSelectBySize(pixt, 6, 32, 8, L_SELECT_IF_BOTH,
L_SELECT_IF_LT, NULL);
count_pieces(pixd, 1);
pixd = pixSelectBySize(pixt, 5, 25, 8, L_SELECT_IF_BOTH,
L_SELECT_IF_GT, NULL);
count_pieces(pixd, 1);
pixd = pixSelectBySize(pixt, 25, 5, 8, L_SELECT_IF_BOTH,
L_SELECT_IF_GT, NULL);
count_pieces(pixd, 1);
pixd = pixSelectByPerimToAreaRatio(pixt, 0.3, 8, L_SELECT_IF_GT, NULL);
count_pieces(pixd, 2);
pixd = pixSelectByPerimToAreaRatio(pixt, 0.15, 8, L_SELECT_IF_GT, NULL);
count_pieces(pixd, 3);
pixd = pixSelectByPerimToAreaRatio(pixt, 0.4, 8, L_SELECT_IF_LTE, NULL);
count_pieces(pixd, 2);
pixd = pixSelectByPerimToAreaRatio(pixt, 0.45, 8, L_SELECT_IF_LT, NULL);
count_pieces(pixd, 3);
pixd = pixSelectByPerimSizeRatio(pixt2, 2.3, 8, L_SELECT_IF_GT, NULL);
count_pieces(pixd, 2);
pixd = pixSelectByPerimSizeRatio(pixt2, 1.2, 8, L_SELECT_IF_GT, NULL);
count_pieces(pixd, 3);
pixd = pixSelectByPerimSizeRatio(pixt2, 1.7, 8, L_SELECT_IF_LTE, NULL);
count_pieces(pixd, 1);
pixd = pixSelectByPerimSizeRatio(pixt2, 2.9, 8, L_SELECT_IF_LT, NULL);
count_pieces(pixd, 3);
pixd = pixSelectByAreaFraction(pixt2, 0.3, 8, L_SELECT_IF_LT, NULL);
count_pieces(pixd, 0);
pixd = pixSelectByAreaFraction(pixt2, 0.9, 8, L_SELECT_IF_LT, NULL);
count_pieces(pixd, 4);
pixd = pixSelectByAreaFraction(pixt2, 0.5, 8, L_SELECT_IF_GTE, NULL);
count_pieces(pixd, 3);
pixd = pixSelectByAreaFraction(pixt2, 0.7, 8, L_SELECT_IF_GT, NULL);
count_pieces(pixd, 2);
boxat = boxaSelectBySize(boxa, 21, 10, L_SELECT_IF_EITHER,
L_SELECT_IF_LT, NULL);
count_pieces2(boxat, 3);
boxat = boxaSelectBySize(boxa, 22, 32, L_SELECT_IF_BOTH,
L_SELECT_IF_LT, NULL);
count_pieces2(boxat, 2);
boxaDestroy(&boxa);
pixDestroy(&pixt);
pixDestroy(&pixt2);
pixDestroy(&pixs);
/* Here's the most general method for selecting components.
* We do it for area fraction, but any combination of
* size, area/perimeter ratio and area fraction can be used. */
pixs = pixRead("feyn.tif");
/* pixs = pixRead("rabi.png"); */
pixc = pixCopy(NULL, pixs); /* subtract bands from this */
pixt = pixCreateTemplate(pixs); /* add bands to this */
pixGetDimensions(pixs, &w, &h, NULL);
boxa = pixConnComp(pixs, &pixas, 8);
n = boxaGetCount(boxa);
fprintf(stderr, "total: %d\n", n);
na1 = pixaFindAreaFraction(pixas);
nat = numaCreate(0);
numaSetCount(nat, n); /* initialize to all 0 */
sum = sumi = 0;
pixac = pixaCreate(0);
for (i = 0; i < 12; i++) {
/* Compute within the intervals using an intersection. */
na2 = numaMakeThresholdIndicator(na1, edges[i], L_SELECT_IF_GTE);
if (i != 11)
na3 = numaMakeThresholdIndicator(na1, edges[i + 1], L_SELECT_IF_LT);
else
na3 = numaMakeThresholdIndicator(na1, edges[i + 1],
L_SELECT_IF_LTE);
na4 = numaLogicalOp(NULL, na2, na3, L_INTERSECTION);
sum += count_ones(na4, 0, 0, NULL);
/* Compute outside the intervals using a union, and invert */
na2i = numaMakeThresholdIndicator(na1, edges[i], L_SELECT_IF_LT);
if (i != 11)
na3i = numaMakeThresholdIndicator(na1, edges[i + 1],
L_SELECT_IF_GTE);
else
na3i = numaMakeThresholdIndicator(na1, edges[i + 1],
L_SELECT_IF_GT);
na4i = numaLogicalOp(NULL, na3i, na2i, L_UNION);
numaInvert(na4i, na4i);
sumi += count_ones(na4i, 0, 0, NULL);
/* Compare the two methods */
if (sum == sumi)
fprintf(stderr, "\nCorrect: sum = sumi = %d\n", sum);
else
fprintf(stderr, "\nWRONG: sum = %d, sumi = %d\n", sum, sumi);
/* Reconstruct the image, band by band. */
numaLogicalOp(nat, nat, na4, L_UNION);
pixad = pixaSelectWithIndicator(pixas, na4, NULL);
pixd = pixaDisplay(pixad, w, h);
pixOr(pixt, pixt, pixd); /* add them in */
pixcount = pixCopy(NULL, pixt); /* destroyed by count_pieces */
count_ones(na4, band[i], i, "band");
count_pieces(pixd, band[i]);
count_ones(nat, total[i], i, "total");
count_pieces(pixcount, total[i]);
pixaDestroy(&pixad);
/* Remove band successively from full image */
pixRemoveWithIndicator(pixc, pixas, na4);
pixSaveTiled(pixc, pixac, 0.25, 1 - i % 2, 25, 8);
numaDestroy(&na2);
numaDestroy(&na3);
numaDestroy(&na4);
numaDestroy(&na2i);
numaDestroy(&na3i);
numaDestroy(&na4i);
}
/* Did we remove all components from pixc? */
pixZero(pixc, &empty);
if (!empty)
fprintf(stderr, "\nWRONG: not all pixels removed from pixc\n");
pixDestroy(&pixs);
pixDestroy(&pixc);
pixDestroy(&pixt);
boxaDestroy(&boxa);
pixaDestroy(&pixas);
numaDestroy(&na1);
numaDestroy(&nat);
/* One last extraction. Get all components that have either
* a height of at least 50 or a width of between 30 and 35,
* and also have a relatively large perimeter/area ratio. */
pixs = pixRead("feyn.tif");
boxa = pixConnComp(pixs, &pixas, 8);
n = boxaGetCount(boxa);
pixaFindDimensions(pixas, &naw, &nah);
na1 = pixaFindPerimToAreaRatio(pixas);
na2 = numaMakeThresholdIndicator(nah, 50, L_SELECT_IF_GTE);
na3 = numaMakeThresholdIndicator(naw, 30, L_SELECT_IF_GTE);
na4 = numaMakeThresholdIndicator(naw, 35, L_SELECT_IF_LTE);
na5 = numaMakeThresholdIndicator(na1, 0.4, L_SELECT_IF_GTE);
numaLogicalOp(na3, na3, na4, L_INTERSECTION);
numaLogicalOp(na2, na2, na3, L_UNION);
numaLogicalOp(na2, na2, na5, L_INTERSECTION);
numaInvert(na2, na2); /* get components to be removed */
pixRemoveWithIndicator(pixs, pixas, na2);
pixSaveTiled(pixs, pixac, 0.25, 1, 25, 8);
pixDestroy(&pixs);
boxaDestroy(&boxa);
pixaDestroy(&pixas);
numaDestroy(&naw);
numaDestroy(&nah);
numaDestroy(&na1);
numaDestroy(&na2);
numaDestroy(&na3);
numaDestroy(&na4);
numaDestroy(&na5);
pixDisplayMultiple("/tmp/display/file*");
pixd = pixaDisplay(pixac, 0, 0);
pixDisplay(pixd, 100, 100);
pixWrite("/tmp/comp.jpg", pixd, IFF_JFIF_JPEG);
pixDestroy(&pixd);
pixaDestroy(&pixac);
return 0;
}
l_int32 main(int argc,
char **argv)
{
l_int32 i, w, h, n, val, ne, no, nbins, minw, maxw, minh, maxh;
l_int32 mine, mino, maxe, maxo;
l_int32 w_diff, h_diff, median_w_diff, median_h_diff;
l_int32 noutw, nouth;
l_float32 medwe, medhe, medwo, medho;
BOXA *boxa1, *boxa2, *boxae, *boxao;
NUMA *na1, *nawe, *nahe, *nawo, *naho;
NUMA *nadiffw, *nadiffh; /* diff from median w and h */
NUMA *naiw, *naih; /* indicator arrays for small outlier dimensions */
NUMA *narbwe, *narbhe, *narbwo, *narbho; /* rank-binned w and h */
PIX *pix1;
PIXA *pixa1;
L_REGPARAMS *rp;
if (regTestSetup(argc, argv, &rp))
return 1;
lept_mkdir("lept/boxa");
boxa1 = boxaRead("boxa4.ba");
/* Fill invalid boxes */
n = boxaGetCount(boxa1);
na1 = boxaFindInvalidBoxes(boxa1);
if (na1)
boxa2 = boxaFillSequence(boxa1, L_USE_SAME_PARITY_BOXES, 0);
else
boxa2 = boxaCopy(boxa1, L_CLONE);
boxaDestroy(&boxa1);
/* Get the widths and heights for even and odd parity */
boxaSplitEvenOdd(boxa2, 0, &boxae, &boxao);
boxaGetSizes(boxae, &nawe, &nahe);
boxaGetSizes(boxao, &nawo, &naho);
boxaDestroy(&boxa2);
/* Find the medians */
numaGetMedian(nawe, &medwe);
numaGetMedian(nahe, &medhe);
numaGetMedian(nawo, &medwo);
numaGetMedian(naho, &medho);
/* Find the median even/odd differences for width and height */
median_w_diff = L_ABS(medwe - medwo);
median_h_diff = L_ABS(medhe - medho);
regTestCompareValues(rp, 210, median_w_diff, 0.0); /* 0 */
regTestCompareValues(rp, 15, median_h_diff, 0.0); /* 1 */
if (rp->display) {
fprintf(stderr, "diff of e/o median widths = %d\n", median_w_diff);
fprintf(stderr, "diff of e/o median heights = %d\n", median_h_diff);
}
/* Find the differences of box width and height from the median */
nadiffw = numaMakeConstant(0, n);
nadiffh = numaMakeConstant(0, n);
ne = numaGetCount(nawe);
no = numaGetCount(nawo);
for (i = 0; i < ne; i++) {
numaGetIValue(nawe, i, &val);
numaSetValue(nadiffw, 2 * i, L_ABS(val - medwe));
numaGetIValue(nahe, i, &val);
numaSetValue(nadiffh, 2 * i, L_ABS(val - medhe));
}
for (i = 0; i < no; i++) {
numaGetIValue(nawo, i, &val);
numaSetValue(nadiffw, 2 * i + 1, L_ABS(val - medwo));
numaGetIValue(naho, i, &val);
numaSetValue(nadiffh, 2 * i + 1, L_ABS(val - medho));
}
/* Don't count invalid boxes; set the diffs to 0 for them */
if (na1) {
for (i = 0; i < n; i++) {
numaGetIValue(na1, i, &val);
if (val == 1) {
numaSetValue(nadiffw, i, 0);
numaSetValue(nadiffh, i, 0);
}
}
}
/* Make an indicator array for boxes that differ from the
* median by more than a threshold value for outliers */
naiw = numaMakeThresholdIndicator(nadiffw, 90, L_SELECT_IF_GT);
naih = numaMakeThresholdIndicator(nadiffh, 90, L_SELECT_IF_GT);
numaGetCountRelativeToZero(naiw, L_GREATER_THAN_ZERO, &noutw);
numaGetCountRelativeToZero(naih, L_GREATER_THAN_ZERO, &nouth);
regTestCompareValues(rp, 24, noutw, 0.0); /* 2 */
regTestCompareValues(rp, 0, nouth, 0.0); /* 3 */
if (rp->display)
fprintf(stderr, "num width outliers = %d, num height outliers = %d\n",
noutw, nouth);
numaDestroy(&nadiffw);
numaDestroy(&nadiffh);
numaDestroy(&naiw);
numaDestroy(&naih);
/* Find the rank bins for width and height */
nbins = L_MAX(5, ne / 50); // up to 50 pages/bin
numaGetRankBinValues(nawe, nbins, NULL, &narbwe);
numaGetRankBinValues(nawo, nbins, NULL, &narbwo);
numaGetRankBinValues(nahe, nbins, NULL, &narbhe);
numaGetRankBinValues(naho, nbins, NULL, &narbho);
numaDestroy(&nawe);
numaDestroy(&nawo);
numaDestroy(&nahe);
numaDestroy(&naho);
/* Find min and max binned widths and heights; get the max diffs */
numaGetIValue(narbwe, 0, &mine);
numaGetIValue(narbwe, nbins - 1, &maxe);
numaGetIValue(narbwo, 0, &mino);
numaGetIValue(narbwo, nbins - 1, &maxo);
minw = L_MIN(mine, mino);
maxw = L_MAX(maxe, maxo);
w_diff = maxw - minw;
numaGetIValue(narbhe, 0, &mine);
numaGetIValue(narbhe, nbins - 1, &maxe);
numaGetIValue(narbho, 0, &mino);
numaGetIValue(narbho, nbins - 1, &maxo);
minh = L_MIN(mine, mino);
maxh = L_MAX(maxe, maxo);
h_diff = maxh - minh;
numaDestroy(&narbwe);
numaDestroy(&narbhe);
numaDestroy(&narbwo);
numaDestroy(&narbho);
regTestCompareValues(rp, 409, w_diff, 0.0); /* 4 */
regTestCompareValues(rp, 49, h_diff, 0.0); /* 5 */
if (rp->display)
fprintf(stderr, "Binned rank results: w_diff = %d, h_diff = %d\n",
w_diff, h_diff);
/* Plot the results */
if (noutw > 0 || nouth > 0) {
pixa1 = pixaCreate(2);
boxaPlotSizes(boxae, "even", NULL, NULL, &pix1);
pixaAddPix(pixa1, pix1, L_INSERT);
boxaPlotSizes(boxao, "odd", NULL, NULL, &pix1);
pixaAddPix(pixa1, pix1, L_INSERT);
pix1 = pixaDisplayTiledInRows(pixa1, 32, 1500, 1.0, 0, 30, 2);
regTestWritePixAndCheck(rp, pix1, IFF_PNG); /* 6 */
pixDisplayWithTitle(pix1, 100, 100, NULL, rp->display);
pixDestroy(&pix1);
pixaDestroy(&pixa1);
}
boxaDestroy(&boxae);
boxaDestroy(&boxao);
return regTestCleanup(rp);
}