Department of Computer Science & Engineering

LAB MANUAL

Program: Computer Science & Engineering

Subject: Advanced Digital Image Processing
Index

Lab 1: Write a program for image enhancement

Lab2: Write a program for image compression
Lab3: Write a program for color image processing
Lab4: Write a program for image segmentation
Lab 5: Write a program for image morphology
Lab 6: Image Restoration
Lab 7: Edge detection
Lab 8: Blurring 8 bit color versus monochrome
Mini Project (Select One)
1. Take a hand written document, Perform preprocessing and try to segment
into characters
2. Take an image, design fuzzy rules for content based image retrieval.
3. Take an image, design a neural network for content based image retrieval.
 LAB 1: Write a program for image enhancement
Adjusting Intensity Values to a Specified Range

You can adjust the intensity values in an image using the imadjust function, where you specify
the range of intensity values in the output image.

For example, this code increases the contrast in a low-contrast grayscale image by remapping the
data values to fill the entire intensity range [0, 255].
I = imread('pout.tif');
J = imadjust(I);
imshow(J)
figure, imhist(J,64)

This figure displays the adjusted image and its histogram. Notice the increased contrast in the
image, and that the histogram now fills the entire range.

Specifying the Adjustment Limits

You can optionally specify the range of the input values and the output values using imadjust.
You specify these ranges in two vectors that you pass to imadjust as arguments. The first vector
specifies the low- and high-intensity values that you want to map. The second vector specifies the
scale over which you want to map them.

For example, you can decrease the contrast of an image by narrowing the range of the data. In the
example below, the man's coat is too dark to reveal any detail. imadjust maps the range [0,51] in
the uint8 input image to [128,255] in the output image. This brightens the image considerably,
and also widens the dynamic range of the dark portions of the original image, making it much
easier to see the details in the coat. Note, however, that because all values above 51 in the original
image are mapped to 255 (white) in the adjusted image, the adjusted image appears washed out.
I = imread('cameraman.tif');
J = imadjust(I,[0 0.2],[0.5
1]); imshow(I)
figure, imshow(J)

Setting the Adjustment Limits Automatically

To use imadjust, you must typically perform two steps: View the histogram of the image to
determine the intensity value limits. Specify these limits as a fraction between 0.0 and 1.0 so that
you can pass them to imadjust in the [low_in high_in] vector.

For a more convenient way to specify these limits, use the stretchlim function. (The imadjust
function uses stretchlim for its simplest syntax, imadjust(I).)
This function calculates the histogram of the image and determines the adjustment limits
automatically. The stretchlim function returns these values as fractions in a vector that you can
pass as the [low_in high_in] argument to imadjust; for example:
I = imread('rice.png');
J = imadjust(I,stretchlim(I),[0 1]);

By default, stretchlim uses the intensity values that represent the bottom 1% (0.01) and the top
1% (0.99) of the range as the adjustment limits. By trimming the extremes at both ends of the
intensity range, stretchlim makes more room in the adjusted dynamic range for the remaining
intensities. But you can specify other range limits as an argument to stretchlim. See the stretchlim
reference page for more information.

Gamma Correction

imadjust maps low to bottom, and high to top. By default, the values between low and high are
mapped linearly to values between bottom and top. For example, the value halfway between low
and high corresponds to the value halfway between bottom and top.

imadjust can accept an additional argument that specifies the gamma correction factor.
Depending on the value of gamma, the mapping between values in the input and output images
might be nonlinear. For example, the value halfway between low and high might map to a value
either greater than or less than the value halfway between bottom and top.

Gamma can be any value between 0 and infinity. If gamma is 1 (the default), the mapping is
linear. If gamma is less than 1, the mapping is weighted toward higher (brighter) output values. If
gamma is greater than 1, the mapping is weighted toward lower (darker) output values.

The figure below illustrates this relationship. The three transformation curves show how values
are mapped when gamma is less than, equal to, and greater than 1. (In each graph, the x-axis
represents the intensity values in the input image, and the y-axis represents the intensity values in
the output image.)

The example below illustrates gamma correction. Notice that in the call to imadjust, the data
ranges of the input and output images are specified as empty matrices. When you specify an
empty matrix, imadjust uses the default range of [0,1]. In the example, both ranges are left empty;
this means that gamma correction is applied without any other adjustment of the data.
[X,map] = imread('forest.tif')
I = ind2gray(X,map);
J = imadjust(I,[],[],0.5);
imshow(I)
figure, imshow(J)
 Lab 2: Write a program for image compression

Aa standard matlab wavelet code package, WAVELAB 802, to perform the transforms.
The wavelets we chose to use were the Deslauriers wavelets of polynomial size 3.

The compression scheme we used was to set a threshold value that was some fraction of the norm
of the entire wavelet transform matrix. If the magnitude of a wavelet in the representation was not
larger than this value, it was not included in the compression. We then rebuilt an image which (to
some degree that depended on how many bases we included) resembed the original image by
running the inverse transform.

ORIGINAL IMAGE
The fabulous Lena
BASIS VECTORS

These are some basis vectors obtained by running the inverse wavelet transform on a matrix
with a single nonzero value in it.

Deslaurier(1,1)

Deslaurier(10,10)

Deslaurier(16,16)
 Deslaurier(4,2)

COMPRESSED IMAGES
Threshold = .5, Bases included = 19, Compression ratio = 3400 : 1
Threshold = 1, Bases included = 470, Compression ratio = 140 : 1

Threshold = 2, Bases included = 2383, Compression ratio = 27 : 1

Threshold = 4, Bases included = 6160, Compression ratio = 10 : 1

Threshold = 8, Bases included = 12378, Compression ratio = 5 : 1

MATLAB CODE
load /home/nirav/elec301/lena256.mat;
imagesc(lena256); colormap(gray(256));

[qmf, dqmf] = MakeBSFilter('Deslauriers', 3);

% The MakeBSFilter function creates biorthonormal filter pairs. The filter

% pairs that we're making is an Interpolating (Deslauriers-Dubuc) filter
% of polynomial degree 3

wc = FWT2_PB(lena256, 1, qmf, dqmf);

% wc correspond to the wavelet coefficients of the sample image

% FWT2_PB is a function that takes a 2 dimensional wavelet transform
% We specify the image matrix, the level of coarseness (1), the quadrature
% mirror filter (qmf), and the dual quadrature mirror filter (dqmf)

% we take a tolerance which is some fraction

% of the norm of the sample image

nl = norm(lena256) / (4 * norm(size(lena256)));

% if the value of the wavelet coefficient matrix at a particular

% row and column is less than the tolerance, we 'throw' it out
% and increment the zero count.

zerocount = 0;
for i = 1:256
for j = 1:256
if ( abs(wc(i,j)) < nl)
 wc(i,j) = 0;
zerocount = zerocount +
1; end
end
end

x = IWT2_PB(wc, 1, qmf, dqmf);

imagesc(x);

% here is some sample code to view how these deslauriers wavelets look

[qmf, dqmf] = MakeBSFilter('Deslauriers',

3); for i = 1:256
for j = 1:256
wc(i,j) = 0;
end
end

% this is the Deslauriers(4,2) matrix

wc(4, 2) = 1000;
x = IWT2_PB(wc, 1, qmf, dqmf);
imagesc(x);
 LAB 3: Write a program for color image processing
Color Approximation

To reduce the number of colors in an image, use the rgb2ind function. This function converts a
truecolor image to an indexed image, reducing the number of colors in the process. rgb2ind
provides the following methods for approximating the colors in the original image: Quantization
Uniform quantization Minimum variance quantization Colormap mapping

The quality of the resulting image depends on the approximation method you use, the range of
colors in the input image, and whether or not you use dithering. Note that different methods work
better for different images. See Dithering for a description of dithering and how to enable or
disable it.

Quantization

Reducing the number of colors in an image involves quantization. The function rgb2ind uses
quantization as part of its color reduction algorithm. rgb2ind supports two quantization methods:
uniform quantization and minimum variance quantization.

An important term in discussions of image quantization is RGB color cube, which is used
frequently throughout this section. The RGB color cube is a three-dimensional array of all of the
colors that are defined for a particular data type. Since RGB images in MATLAB can be of type
uint8, uint16, or double, three possible color cube definitions exist. For example, if an RGB
image is of class uint8, 256 values are defined for each color plane (red, blue, and green), and, in
total, there will be 224 (or 16,777,216) colors defined by the color cube. This color cube is the
same for all uint8 RGB images, regardless of which colors they actually use.

The uint8, uint16, and double color cubes all have the same range of colors. In other words, the
brightest red in a uint8 RGB image appears the same as the brightest red in a double RGB image.
The difference is that the double RGB color cube has many more shades of red (and many more
shades of all colors). The following figure shows an RGB color cube for a uint8 image.

Quantization involves dividing the RGB color cube into a number of smaller boxes, and then
mapping all colors that fall within each box to the color value at the center of that box.

Uniform quantization and minimum variance quantization differ in the approach used to divide
up the RGB color cube. With uniform quantization, the color cube is cut up into equal-sized
boxes (smaller cubes). With minimum variance quantization, the color cube is cut up into boxes
(not necessarily cubes) of different sizes; the sizes of the boxes depend on how the colors are
distributed in the image.

Uniform Quantization. To perform uniform quantization, call rgb2ind and specify a tolerance.
The tolerance determines the size of the cube-shaped boxes into which the RGB color cube is
divided. The allowable range for a tolerance setting is [0,1]. For example, if you specify a
tolerance of 0.1, the edges of the boxes are one-tenth the length of the RGB color cube and the
maximum total number of boxes is
n = (floor(1/tol)+1)^3

The commands below perform uniform quantization with a tolerance of 0.1.

RGB = imread('peppers.png');
[x,map] = rgb2ind(RGB, 0.1);

The following figure illustrates uniform quantization of a uint8 image. For clarity, the figure
shows a two-dimensional slice (or color plane) from the color cube where red=0 and green and
blue range from 0 to 255. The actual pixel values are denoted by the centers of the x's.

After the color cube has been divided, all empty boxes are thrown out. Therefore, only one of the
boxes is used to produce a color for the colormap. As shown earlier, the maximum length of a
colormap created by uniform quantization can be predicted, but the colormap can be smaller than
the prediction because rgb2ind removes any colors that do not appear in the input image.

Minimum Variance Quantization. To perform minimum variance quantization, call rgb2ind and
specify the maximum number of colors in the output image's colormap. The number you specify
determines the number of boxes into which the RGB color cube is divided. These commands use
minimum variance quantization to create an indexed image with 185 colors.
RGB = imread('peppers.png');
[X,map] = rgb2ind(RGB,185);

Minimum variance quantization works by associating pixels into groups based on the variance
between their pixel values. For example, a set of blue pixels might be grouped together because
they have a small variance from the center pixel of the group.

In minimum variance quantization, the boxes that divide the color cube vary in size, and do not
necessarily fill the color cube. If some areas of the color cube do not have pixels, there are no
boxes in these areas.

While you set the number of boxes, n, to be used by rgb2ind, the placement is determined by the
algorithm as it analyzes the color data in your image. Once the image is divided into n optimally
located boxes, the pixels within each box are mapped to the pixel value at the center of the box,
as in uniform quantization.

The resulting colormap usually has the number of entries you specify. This is because the color
cube is divided so that each region contains at least one color that appears in the input image. If
the input image uses fewer colors than the number you specify, the output colormap will have
fewer than n colors, and the output image will contain all of the colors of the input image.

The following figure shows the same two-dimensional slice of the color cube as shown in the
preceding figure (demonstrating uniform quantization). Eleven boxes have been created using
minimum variance quantization.
For a given number of colors, minimum variance quantization produces better results than
uniform quantization, because it takes into account the actual data. Minimum variance
quantization allocates more of the colormap entries to colors that appear frequently in the input
image. It allocates fewer entries to colors that appear infrequently. As a result, the accuracy of the
colors is higher than with uniform quantization. For example, if the input image has many shades
of green and few shades of red, there will be more greens than reds in the output colormap. Note
that the computation for minimum variance quantization takes longer than that for uniform
quantization.

Colormap Mapping

If you specify an actual colormap to use, rgb2ind uses colormap mapping (instead of
quantization) to find the colors in the specified colormap that best match the colors in the RGB
image. This method is useful if you need to create images that use a fixed colormap. For example,
if you want to display multiple indexed images on an 8-bit display, you can avoid color problems
by mapping them all to the same colormap. Colormap mapping produces a good approximation if
the specified colormap has similar colors to those in the RGB image. If the colormap does not
have similar colors to those in the RGB image, this method produces poor results.

This example illustrates mapping two images to the same colormap. The colormap used for the
two images is created on the fly using the MATLAB function colorcube, which creates an RGB
colormap containing the number of colors that you specify. (colorcube always creates the same
colormap for a given number of colors.) Because the colormap includes colors all throughout the
RGB color cube, the output images can reasonably approximate the input images.
RGB1 = imread('autumn.tif');
RGB2 = imread('peppers.png');
X1 = rgb2ind(RGB1,colorcube(128));
X2 = rgb2ind(RGB2,colorcube(128));
 Lab 4: Write a program for image segmentation
demonstration of global and local thresholding for segmentation

% threshdemo.m
% Demonstration of global and local threshold operations of an image

clear all
[tmp,idx]=imread('lena.bmp');
a=ind2gray(tmp,idx); % gray scale image of lena, value between 0 and 1
clear tmp idx
figure(1),clf,colormap('gray'),imshow(a),title('original Lena
image') [m,n]=size(a); % size of image a
b=reshape(a,m*n,1); % into a column vector
figure(2),hist(b,50),title('histogram of image')
% first do global thresholding
mu=rand(2,1); % value betwen 0 and 1, two clusters only
[W,iter,Sw,Sb,Cova]=kmeansf(b,mu);% W is the mean,
% Cova is the covariance matrices
% member: membership of each X: K by 1 vector of elements 1 to c
[d,member]=kmeantest(b,sort(W));
c=reshape(member-1,m,n);
clear d member b
figure(3),clf,colormap('gray'),imshow(c)
title('global threshold')

% next do local threshold, partition the image into 64 x 64 blocks

% and do threshold within each block
c=zeros(512,512); trials=0;
for i=1:8,
for j=1:8,
trials=trials+1;
disp([int2str(trials) ' of 64 iterations ...']);
tmp=a(64*(i-1)+1:64*i,64*(j-
1)+1:64*j); tmpi=reshape(tmp,64*64,1);
mu=sort(rand(2,1)); % value betwen 0 and 1, two clusters only
[W,iter,Sw,Sb,Cova]=kmeansf(tmpi,mu);% W is the mean,
% Cova is the covariance matrices
% member: membership of each X: K by 1 vector of elements 1 to c
[d,member]=kmeantest(tmpi,sort(W)); c(64*(i-1)+1:64*i,64*(j-
1)+1:64*j)=reshape(member,64,64);
end
end
figure(4),clf,colormap('gray'),imshow(c-
1), title('local threshold, 64x64 block');
 Lab 5: Write a program for image morphology

demonstrate boundary extraction, interior filling

% demonstrate morphological boundary extraction

%
clear all, close all

A0=imread('myshap4.bmp');
imshow(A0); % a heart shape hand drawing
title('original image');
pause
% A0 contains mostly 1s and the drawing contains 0s, uint8
A1=1-double(A0); % invert black and white

B=ones(3);
A2=imopen(imclose(A1,B),B); % fill 1 pixel hole and remove sticks

A3=imfill(A2,[100 100]); % fill the interior

A=double(A2) + double(A3);

imshow(A),title('after interior filling using

imfill'); pause
Ab=A-double(erode(A,B));

imshow(Ab), title('extracted boundary');

clear A1 A2 A3; Ac=Ab;

vidx=[[1:20:280] 280];

Ac(vidx,:)=1; Ac(:,vidx)=1;

imshow(Ac)
Lab 6: Program for Image Restoration
% load image
X = double(imread('midterm.bmp'));
X = X-mean(X(:));
[m,n] = size(X);

% show image and

DFT fX = fft2(X);
figure(1)
imshow(real(X),[]);
title('original
image') figure(2)
imshow(fftshift(log(1+abs(fX))),[])
title('log(1+|DFT|)) original image');

% model blurring
filter s = 24; t= 0;
u = 1; v=0;
g = [ones(s,1);zeros(m-s-t,1); ones(t,1)];
%g = [ones(s,1);0.99; zeros(m-s-t-2,1);0.99;
ones(t,1)]; g = g/sum(abs(g));
h = [ones(u,1); zeros(n-u-v,1);
ones(v,1)]; h = h/sum(abs(h));
f =g*h';
ff = fft2(f);
figure(3)
imshow(fftshift(log(1+abs(ff))),[])
title('amplitude: log(1+|OTF|)');
figure(4)
imshow(fftshift(angle(ff)),[])
title('phase of OTF');

% get pseudo inverse filter

ff(find(abs(ff)==0))=NaN;
aff = abs(ff);
pff = ff./aff;
apiff = 1./aff;
ppiff = conj(pff);
ppiff(find(isnan(ppiff))) =
0; cap = 11;
apiff(find(apiff > cap)) =
cap; apiff(find(isnan(apiff)))
= 0; piff = apiff.*ppiff;

% deblur and show

frX = piff.*fX;
rX =real(ifft2(frX));
figure(5)
imshow(fftshift(log(1+abs(frX))),[])
title('log(1+|DFT|)) restored
image') figure(6)
imshow(rX(:,5:n),[]);
title('restored image')
Lab 7: Program for Edge detection
% demonstrate edge detection

% numbers of
colors sncols=128;
ncols=32;

% get image from MATLAB

image load('trees');

% show original image

figure(1);
showgimg(real(X),sncols);
drawnow;

% construct convolution
functions [m,n] = size(X);
gs = [1 -1]; ge = [];
hs = [1 -1]; he = [];
g = [gs,zeros(1,m-length(gs)-length(ge)),ge];
h = [hs,zeros(1,n-length(hs)-length(he)),he];

% construct convolution matrices as sparse

matrices Y = spcnvmat(g);
Z = spcnvmat(h);
Wg = Y*X;
Wh = X*Z';

% show transformed images

figure(2);
showgimg(Wg,ncols);
drawnow;

figure(3)
showgimg(Wh,ncols);
drawnow;

figure(4)
showgimg(abs(Wg)+abs(Wh),ncols);
drawnow;
Lab 8: Blurring 8 bit color versus monochrome

% smoothing in eight bit color and monochrome

% parameter definition

% get image from MATLAB

library load('clown');

% construct convolution functions

[m,n] = size(X);
gs = [0.5 0.5]; ge = [];
hs = [0.5 0.5]; he = [];
g = [gs,zeros(1,m-length(gs)-length(ge)),ge];
h = [hs,zeros(1,n-length(hs)-length(he)),he];

% construct convolution matrices and blur sparse

matrices Y = spcnvmat(g);
Z = spcnvmat(h);
W = Y*X*Z';

% show original and blurred images

figure(1);
imshow(X,[]);
figure(2);
imshow(W,[]);
figure(3);
imshow(X,[]);
colormap(map)
figure(4);
imshow(W,[]);
colormap(map)
 Mini Project

Segmentation in handwritten document

Horizontal and Vertical Histogram

Horizontal and Vertical Histogram technique is one the handwritten script identified technique which uses
the Matra/Shirorekha based feature, In this technique longest horizontal run of black pixels on the rows of
a Bangla text word will be much longer than that of English script. This is so because the characters in a
Bangla word are generally connected by matra/Shirorekha (see Fig.2). Here row-wise histogram of the
longest horizontal run is shown in the right part of the words. This information has been used to separate
English from Bangla script. Matra feature is considered to be present in a word, if the length of the
longest horizontal run of the word satisfies the following two conditions: (a) if it is greater than 45% of
the width of a word, and (b) if it is greater than thrice of the height of busy-zone.

Fig. 2. The matra/Shiroreka feature in Bangla and English words

Curvature Information and Local Extreme of Curvature

Curvature information gives a unique, viewpoint independent description for local shape. In
differential geometry, it is well known that a surface can be reconstructed up to second order (except for a
constant term) if the two principal curvatures at each point is known, by using the first and second
fundamental forms Therefore, curvature information provides a useful shape descriptor for various tasks in
computer vision, ranging from image segmentation and feature extraction, to scene analysis and object
recognition . In the field of handwritten alphanumeric character recognition , many researcher use
difference method for recognition such as geometrical and topological feature , statistic feature , and other
algorithms to perform recognition based on character shape. But for a good handwritten recognition system
depends on main two attributes, first selected feature gathering from a handwritten character, second the
recognizers that trained to remember feature of each character in order to cluster and recognize each input
character. In curvature information handwritten character is used as a sequence of curve segments. Each
curve segment is characterized by its degree of curvature which is measure by the cumulative angle
difference from all sampling points with in the segments. If the cumulative is minus, it is a clockwise
curve. Since some characters may consist of the same number of segments with the same curve (ex. one
segment with clockwise curve) other features of segments are gathering to distinguish thes
characters.

Topological Features

Topological features such as loops is a group of white pixels surrounded by black ones ,end points is
point with exactly 1 neighboring point ,dots a cluster of say 1-3 pixels and junction is a point with more
than 2 neighbors all in thinned black and white images are shown in fig 3.

Fig.3. Loop, End points, a Dot and Junctions.

Dots: Dots above the letters “i” and “j” can be identified with a simple set of rules. Short, isolated strokes
occurring on or above the half-line are marked as potential dots.

Junctions: Junctions occur where two strokes meet or cross and are easily found in the skeleton as points
with more than two neighbors.

Endpoints: Endpoints are points in the skeleton with only one neighbor and mark the ends of strokes,
though some are artifacts of the skeletonization algorithm.

Loops: Loops are found from connected-component analysis on the smoothed image, to find areas of
background color not connected to the region surrounding the word. A loop is coded by a number
representing its area.

Parameters of polynomials
Recognition for handwritten mathematical expressions is more complicated than for ordinary handwriting
because of the two-dimensional nature of the input, because there are more symbols involved, and because
the greater difficulty of using contextual information (e.g. harder to provide a dictionary of words or
phrases against which the input is probably coming from). Since past techniques have not provided a
complete solution. To overcome this problem we use parameters of polynomial as one of the technique for
feature extraction. Here we take the idea of Chebyshev polynomials. Using a collection of writing samples
of mathematical symbols, we find that in many cases it provides a succinct way to model the stylus
movements of actual test users. The principal contributions of this technique are:
• To find that modeling x(t) and y(t) as Chebyshev series accurately captures the shape of handwritten
mathematical characters using few parameters.
• To find the polynomial coefficients from the writing samples form clusters which often contain the same
character as written by different test users.
For our analysis, we assume that handwriting traces are provided as sequences of (xi, yi, ti) tuples that
have been normalized so t0 = 0 and tn = 1. The (x,y) trace of character is shown in figure 4. For basis
functions we use Chebyshev polynomials of the first kind, defined by T n(t) ≡ cos(n arccos t) and
orthonormal for the inner product (f, g ) = ∫ f(t)g(t)dt with limit (-1,1)t. The Chebyshev series for X and Y
is

∞
X(t) = ∑ αiTi(t)
i=0

∞
Y(t) = ∑ βiTi(t)
i=0

The (x, y) trace of G

Contour Information
Given a binary image, it is scanned from top to bottom and right to left, and transitions from white
(background) to black (foreground) are detected. The contour is traced counterclockwise outside the pattern
(clockwise for interior contours) and expressed as an array of contour elements shown in figure 5. Each
contour element represents a pixel on the contour and contains fields for the x,y coordinates of the pixel,
the slope or direction of the contour into the pixel, and auxiliary information such as curvature.
Techniques that are useful in recognition of words are described in this subsection: (i) Determination of
upper and lower contours of the word, (ii) Determination of significant local extrema on the contour, (iii)
Determination of reference lines, and (iv) Determination of word length.

1. Upper and lower contours: Division of an exterior contour into upper and lower parts (Figure 6) involves
the detection of two “turnover” points on the contour - the points at which the lower contour changes to upper
contour and vice versa. Given that the exterior contour is traced in the counterclockwise direction, the upper
contour runs predominantly from right to left and the lower predominantly from left to right. The left end of the
leading ligature and the right end of the ending ligature constitute turnover points.

Splitting a contour to upper and lower parts. The upper contour is shown in
dotted lines and the lower contour is shown in bold lines
2. Local Contour Extrema: Local extrema represent a powerful abstraction of the shape of the word.
Local Y-extrema is simply derived from the y-coordinates of the pixels on the chaincoded contour. The
primary challenge is the detection and rejection of spurious extrema arising from irregularities in the
contour. Heuristic filters are used to eliminate spurious extrema. These include checks on the distance of
the detected extremum from previous extrema detected, spatial postion relative to other extrema, the slope
of the contour in the neighborhood of the extremum, and so forth. Majority of spurious extrema in discrete
writing are due to ends of strokes used to form characters, and generally occur in the middle zone of the
word. Fragmentation in the binary image due to suboptimal binarization or poor resolution often leads to
broken strokes and spurious extrema. Since the upper contour flows effectively from right to left, left to
right excursions of the contour must be followed by right to left retraces, and this leads to additional
extrema being detected. Extrema found on such uncharacteristic excursions must be discarded lest they
lead to spurious ascenders. A similar problem arises with right-to-left stretches of the lower contour in the
context of spurious descenders. Let us refer to extrema on such uncharacteristic excursions of the upper
and lower contours as being directionally invalid. Directional validity is conveniently determined from the
slopes of the contour elements into and out of the extremum and is sufficient to discard most spurious
extrema.

3. Reference Line:
A time-tested technique for global reference line determination is based on the vertical histogram of pixels
. The method makes the implicit assumption that the word was written horizontally and scanned in without
skew. Consequently, the method fails for words with significant baseline skew. Unfortunately, baseline
skew is present to varying degrees in most freeform handwriting. Baseline skew can be corrected by
estimating the skew of the word and applying a rotation or shear transformation on the image. An
alternative to skew correction is the use of reference lines of the form r i(x)= m *x+c where m is the skew
angle and c is the offset of the corresponding reference line from the x-axis. Angled reference lines may be
computed from the angular histogram at the skew angle.

The best-fit line through the minima points may be determined by a least-square linear regression
procedure (Figure 7). Minima that do not fall in the vicinity of the implicit baseline either correspond to
descenders, or are spurious. The halfline may be determined as the regression line through upper contour
maxima. However upper contour maxima are often poorly aligned, and spurious points are difficult to
detect and remove, especially in more discrete writing (non-cursive) styles. Consequently the resulting
halfline is often erroneous. Reference lines computed from the angular histogram at the skew angle of
the baseline have proved to be more reliable (Figure 7(c)).

a) b) c)

. (a) Exterior contour of word showing lower-contour minima. (b) Baseline determined as regression
line through minima. (c) Angular reference lines from angular histogram at skew angle.

4. Word Length, Ascenders, and Descenders: The number of minima on the lower contour of a word
image is a good estimate of the length of a word. It is expected that such a length would be proportional to
the number of letters in the word. Figure 4 shows the length estimate of the word as 11 and also shows one
ascender and one descender. The positions of ascsenders and descenders along with the length of the word
are used as features. An ascender refers to the portion of a letter in a word that rises above the general
body of the word. A descender refers to the portion of the word that falls below the general body of the
word. Ascenders and descenders can be appropriately selected from the extremas of the upper and lower
contour of the word.

Local contour extrema: maximas and minimas on the exterior and interior
contours are marked along with the reference lines. The loop associated with
‘b’ that rises above the reference lines in the middle is called an ascender and the
loop associated with the ’q’ that falls below the reference lines in the middle is
called a descender.