1. Images and Programming
Basic MATLAB programming
Grayscale Images
RGB Images
Indexed Color Images
Data Types and Conversions
Image Files and Formats

2. MATLAB Basic Use of MATLAB Where to find more helps?
Textbook: Appendix A
Where to find more helps?

3. Introduction to MATLAB
MATLAB is a data analysis and visualization tool
It has been designed with powerful support for matrices and matrix operations.
It has excellent graphics capabilities and its own powerful programming language.
There are sets of MATLAB programs designed to support particular tasks called toolboxes
Image Processing Toolbox

4. Introduction to MATLAB
MATLAB’s standard data type is the matrix, all data are considered to be matrices of some sort.
Images are matrices whose elements are the gray values or RGB values of its pixels.
A single value is considered to be a 1x1 matrix
A string is a 1xn matrix of characters where n is the string’s length

5. Introduction to MATLAB
When you start up MATLAB, the MATLAB desktop will appear as shown in the Figure.
The prompt consists of two right arrows: >>

6. Basic Use of MATLAB >> 2+2
MATLAB is command line driven; all commands are entered by typing them after the prompt symbol.
>> 2+2
ans = 4

7. Basic Use of MATLAB
MATLAB does all its calculations internally to double precision.
The default display format is to use only four decimal places.
We can change this by using the format function.
>> 11/7
1.5714
>> format long
ans =
Entering the command format returns to the default format

8. Basic Use of MATLAB
MATLAB has all the elementary mathematical functions built in: sqrt, sin, log, log10, pi, etc.

9. Variables and the Workspace
Variables are used to store values.
>> a = 5^(7/2)
a =
>> log(a^2)/log(5)
ans = 7

10. Variables and the Workspace
It lists all your currently defined variables, their numeric data types, and their sizes in bytes. The same information can be obtained using the whos function
>> whos
Name Size Bytes Class
a 1x double array
ans 1x double array
Grand total is 2 elements using 16 bytes

11. Variables and the Workspace
Workspace (cont)
Note that ans is a variable. It is automatically created by MATLAB.
A listing of the variable names is obtained using who function:
>> who
Your variables are
a ans
MATLAB’ standard data type for numbers is double, and they are stored as double-precision 8-byte values.

12. Dealing with Matrices
MATLAB has an enormous number of commands for generating and manipulating matrices.
We can use some of these commands to investigate aspects of the image.
We can enter a small matrix by listing its elements row by row, using spaces or commas as delimiters for the elements in each row and semicolons to separate the rows, for example,
>> a = [ ; ; ; ]

13. Dealing with Matrices Matrix Elements >> a(2, 3) ans = -3
Matrix elements can be obtained by using the standard row, column-indexing scheme.
>> a(2, 3)
ans = -3
Returns the element of the matrix in row 2 and column 3.
MATLAB allows matrix elements to be obtained using a single number; this number being the position where the matrix is written out as a single column. Thus, the order of elements in a is
a(2,3) = a(10) = -3

14. Dealing with Matrices
In general, for a matrix M with r rows and c columns, element m(i, j) corresponds to m(i + r*(j - 1)).
To obtain a row of values, or block of values, we use MATLAB’s colon operator (:), for example,
>> 2:3:16
ans =

15. Dealing with Matrices >> a(3,1:3) Apply to the matrix a, ans =
>> a(2:4,3) -3 -5
>> a(2:3,3:4)
-3 2
>> a(3,:)

16. Dealing with Matrices ans = -2 -5 -8 3 4 1 6 -7 . 7 2 -6

17. Dealing with Matrices Matrix Operations >> flipud(a)
All the standard operations are supported (add, subtract, multiply, invert matrix, matrix power, transpose)
>> 2*a-3*b
>> a^3*b^4
>> inv(a)
>> a’
MATLAB supports some geometric operations on matrices; flipud - flip a matrix up and down, fliplr - flip a matrix left and right., rot90 rotates a matrix by 90 degree.
>> flipud(a)
>> fliplr(a)
>> rot90(a)
ans =

18. Dealing with Matrices >> reshape([1:20],5,4) ?
The reshape function produces a matrix with elements taken column by column from the given matrix.
>> c=[ ; ; ; ]
C =
>> reshape(c,2,10)
>> reshape(c,5,4)
Reshape produces an error if the product of the two values is not equal to the number of elements of the matrix.
>> reshape([1:20],5,4) ?
>> reshape([1:20],5,4)’ ?

19. Dealing with Matrices Array and Matrix arithmetic operators
Dot operator
= Array operator
Note:
1/0 return Inf

20. Dealing with Matrices Vectorization Special Matrices
zeros(m,n) produces a zeros matrix of size m by n
ones(m,n) produces an ones matrix pf size m by n
Vectorization
Refers to an operation carried out over entire matrix or vector
>> tic, for i=1:10^6, sin(i); end, toc
elapsed_time =
>>tic, i=1:10^6; sin(i); toc
elasped_time =
1.3522
Vectorized commands perform very quickly. They should be used instead of for loop.

21. Plots MATLAB has outstanding graphics capabilities.
The plot function can be used to produce many different plots.
>> plot(x,sin(x))
>> plot(x, sin(x), ‘.’, x, cos(x), ‘o’);

22. Help in MATLAB
MATLAB comes with a vast amount of online help and information.
To obtain information on a particular command, you can use help.
>> help lookfor
>> doc help
>> lookfor exp

23. Programming in MATLAB
MATLAB has a small set of built-in functions, others are written in MATLAB’s own programming language.
We consider two distinct programs
Script files
Functions
Script file
It is a list of commands to be executed
Function
It takes an input (one or several variables) and returns one or several values.

24. MATLAB: Load Image Command: imread Syntax: X = imread(‘Filename.ext’);
X : 8-bit/16-bit unsigned int matrix
[X, MAP]= imread(‘Filename.ext’);
MAP: color palette (Type: Double, Range: [0,1])
Supported format: BMP, GIF, JPEG, PBM, PCX, PMG, PNG, PNM, PPM, RAS, TIFF, …

25. MATLAB: Display Image Command: imshow Syntax (not completed):
imshow(‘filename.ext’)
imshow(X)
imshow(X, MAP)
Show pixel value by command pixval on. (Format: column, row = pixel)

26. MATLAB: Write Image Command: imwrite Syntax:
imwrite(X, ‘filename.ext’)
imwrite(X, MAP, ‘filename.ext’)
imwrite(X, ‘filename’, format)
imwrite(X, ‘filename.ext’, param1, value1, param2, value2, …)

27. Example: Load Grayscale Image
>>
w = imread(‘wombats.tif’) ;
Press ENTER and finish..
What happened with this command?
intensities of wombats.tif image is downloaded to matrix w.
size of the matrix w = height  width
No single quote (‘) if the filename is stored in variable

28. Example: Display Grayscale Image
>>
figure
>> imshow(w)
>> pixval on
What happened with these commands?
create figure window (figure command)
display image inside matrix w (imshow command)
show intensity of the pixel (pixval on command)

29. Example: Display Grayscale Image (cont)
figure
>> imshow(‘wombat.tif’)
>> pixval on
What happened with these commands?
create figure window (figure command)
display wombat.tif image (imshow command)
show intensity of the pixel (pixval on command)
No image stored in memory!!!
>>

30. Example: Display Grayscale Image (cont)
Figure 2.1 The wombats image with pixel on.

31. Example: Load RGB Image
>>
a = imread(‘lily.tif’) ;
Press ENTER and finish..
What happened with this command?
color of lily.tif image is downloaded to matrix a.
size of the matrix a = height  width  3
>> size(a)
ans =
index of a: a(row, column, page)
page: 1 = R, 2 = G, 3 = B

32. RGB Color Cube
(a)
(b)

33. Example: Display Pixel Value
>>
impixel(a, 200, 100)
What happened with this command?
show pixel value of image a at column 200 and row 100.
If a is the color image, show R G B values at position (100,200).
If a is the grayscale image, show I I I where I is the intensity at position (100,200).

34. Example: Display Pixel Value
>> inshow(a)
>> impixel(a,88,137)
ans =
>> a(100,200,2)
>> a(100,200,1:3)
>> a(100,200,:)
Figure 2.2 The lily flower image with pixel on.

35. Example: Load Indexed Image
>> em = imread(‘emu.tif’);
>> imshow(em);
Correct?

36. Example: Load Indexed Image
>> em = imread(‘emu.tif’);
We get something wrong. Why?
em = index of emu.tif image
relationship between index and color?

37. Example: Load Indexed Image
>>
[em, emap] = imread(‘emu.tif’);
Press ENTER.
What happened with this command?
index of emu.tif image is downloaded to matrix em.
size of the matrix em = height  width
palette of emu.tif image is downloaded to matrix emap.
size of the matrix emap = 256  3

38. Example: Display Indexed Image
>> figure
>> imshow(em, emap)
>> pixval on
What happened with these commands?
create figure window (figure command)
display color from index em with the palette emap (imshow command)
show value of the pixel (pixval on command)

39. Example: Display Indexed Image

40. Example: Display Indexed Image
Display without color map
Display with color map

41. MATLAB: Image Information
Command: imfinfo
Syntax: imfinfo(‘filename.ext’)
Information shown:
Filename, FileModDate, FileSize, Format, FormatVersion, Width, Height, BitDepth(no of bits per pixel), ColorType (truecolor, grayscale, or indexed), etc

42. MATLAB: Data Types Data Type Description Range
int bit integer to 127
uint bit unsigned integer to 255
int bit integer to 32767
uint bit unsigned integer to 65535
double Double precision real number machine specific
Note: Arithmetic operation allowed only for the data type double.

43. MATLAB: Data Conversion
>> b = uint8(a);
>> b
b = 23
>> whos a b
Name Size Bytes Class
a 1x double array
b 1x uint8 array
Don’t forget to convert the image to double data before performing any computations.

44. MATLAB: Image Conversion
Function Use Format
ind2gray indexedgrayscale y = ind2gray(x,map);
gray2ind grayscaleindexed [y,map] =gray2int(x);
rgb2gray RGBgrayscale y = rgb2gray(x);
gray2rgb grayscaleRGB x = gray2rgb(y);
rgb2ind RGBindexed [x,map] =rgb2int(y);
ind2rgb indexedRGB y = ind2rgb(x,map);

45. Vector Image VS Raster Image
Image = collection of line
Scale up without loss of sharpness
Not suitable for natural screen
E.g. Adobe PostScript
Image = collection of point
Popular for image files
Recommended for image processing
E.g. PGM, BMP, JPEG, …

46. Microsoft BMP Format Header (RAW format, unreadable) BM……….
Header format (54 bytes)
File header: (Bytes 0-13)
Byte Signature BM (42 4D)
Byte 2-5 FileSize File size in Byte
Byte 6-9 Reserved All zeros
Byte DataOffset File offset to raster data

47. Microsoft BMP Format Information header (Byte 14-53)
Byte Size Size of information header
Byte Width Image width [pixel]
Byte Height Image height [pixel]
Byte Planes Number of image plane (=1)
Byte BitCount Bit/Pixel (1, 4, 8, 16, 24)
Byte Compression Type of compression
0 : no compression,
1, 2: RLE encoding
(rarely used)

48. Microsoft BMP Format Byte 34-37 ImageSize Image size
Byte HorizontalRes Horizontal resolution [pixel/meter]
Byte VerticalRes Vertical resolution [pixel/meter]
Byte ColorsUsed #color used in the
image (0=allcolor,2BitCount)
Byte ImportantColors #important color

49. Number Format in BMP Format: Least-endian (little-endian)
LSB on the lowest address
Ex. Byte from BMP file (width):
 actual number

50. GIF Format
Data are compressed by using LZW (Lempel-Ziv-Welch) compression (lossless compression)
Allowed a maximum 256 colors
Not allowed for grayscale/binary images
Allow multiple images in one file (can create animated GIFs)
It is one of the standard formats supported by the WWW
Header
signature (3 bytes): GIF
version (3 bytes): 87a or 89a

51. GIF Format Alternative format: PNG screen width (2 bytes) : unsigned
screen height (2 bytes) : unsigned
etc.
Alternative format: PNG
non-patented algorithm (zlib compression)
also support binary, grayscale, truecolor and indexed images
possible to include alpha channel, gamma correction, …

52. JPEG Format Lossy compression Header: Byte 0-1 Signature FF D8 (HEX)
Byte 2-3 Application Maker FF E0 (HEX)
Byte 4-5 Segment length
Byte JFIF\ 0 ASCII JFIF.
Byte JFIF version Version
Byte 13 Units 0: arbitrary,
1: pixel/inch,
2: pixel/cm.

53. JPEG Format Byte 14-15 Horizontal pixel density
Byte Vertical pixel density
Byte 18 Thumbnail width 0 : no thumbnail
Byte 19 Thumbnail height 0 : no thumbnail

54. TIFF Format Be able to store multiple image files
Allow different compression routines (LZW, JPEG, Huffman, RLE…)
Allow different byte ordering (little or big-endian)
Allow binary, grayscale, indexed, truecolor images
Good for data exchange

55. TIFF Format Header: 8 bytes only
Byte 0-1 Byte order 4D 4D: ASCII MM for big endian
49 49: ASCII II for little endian
Byte 2-3 TIFF version Always 42 (00 2A: big endian,
2A 00: little endian)
Byte 4-8 Image offset Pointer to the position of the data
for the first image

56. DICOM Format Format of medical images Able to store multiple images
Allow for describing 3D images
Header (Very long and variable length!!):
Byte 0-127: preamble (not used)
Byte : signature (DICM)

57. DICOM Format Info in Header
image information (size, #slices, modality used: CAT, MRI, …)
patient information (name, patient ID, …)
compression used (JPEG, RLE, …)
Complex!!! 

58. Example of MATLAB function
function dumphex(filename, n) %
% DUMPHEX(FILENAME, N) prints the first 16*N bytes of the file FILENAME
% in hex and ASCII. For example:
% dumphex('picture.bmp’, 4)
fid = fopen(filename,'r');
if fid == -1
error('File does not exist or is not in your Matlab path');
end;
a=fread(fid,16*n,'uchar');
idx=find(a>=32 & a <=126);
ah=dec2hex(a);
b=repmat([' '], 16*n, 3);
b2=repmat('.', 16, n);
b2(idx)=char(a(idx));
b(:,1:2)=ah;
[reshape(b', 48, n)' repmat(' ', n, 2) reshape(b2, 16, n)']