[Ch1] Introduction

1.1 Images and pictures

Image : a single picture which represents something

 

1.2 What is Image Processing

Image Processing involves changing the nature of an image in order to either

1) improve its pictorial information for human interpretation,

2) render it more suitable for autonomous machine perception.

 

Digital Image Processing involves using a computer to change the nature of a digital image

A procedure which satisfies condition 1) may be the very worst procedure for satisfying condition 2).

Humans like their images to be sharp, clear and detailed; machines prefer those to be simple and uncluttered.

 

1) Improving pictorical information for human interpretation

(1) Enhancing (Sharpening) Edges

 

(2) Removing Noise

: Each type of noise requiring a different method of removal

Noise : Random errors in the image / Very common problem in data transmission

 

(3) Removing Motion Blur

Motion blur : occured when the shutter speed of the camera is too long for the speed of the object

 

2) Rendering it more suitable for autonomous machine perception

(1) Edge detection

- Necessary for measurement of objects

- A first step in edge enhancement

(necessary to enhance the original img slightly, to make edges clearer)

 

(2) Removing Detail by Blurring

For Measurement (size, shape) or Counting purposes, not Interested in all the fine detail in an img

→ Fine detail is gone & Coarse structure of img is remained

 

1.3 Images and Digital Images

1) Image (=Photo)

- Black and white (No colored)

- A two dimensional function, where the function values give the Brightness of the img at any given point

- Real Brightness values bw 0.0 (black) to 1.0 (white)

- The ranges of x and y depend on img, but they can take all real values bw their min and max

 

2) Digital Image

- x, y, and f(x,y) values are all Discrete (Integer values)

- Integer Brightness values bw 0 (black) to 255 (white)

- img has x and y bw 1 to 256 each

- A large array of discrete dots (Pixels), each of which has a brightness associated with it

- The pixels surrounding a given pixel constitute its Neighborhood

Neighbourhood can be characterized by its shape in the same way as a matrix:

3x3 neighbourhood, 5x7 neighbourhood

 

Odd numbers of rows and columns to ensure the current pixel is in the center of the neighborhood

If Even number of rows or columns (or both)?

→ Necessary to specify which pixel in the neighborhood is “current pixel”

 

 

 

1.4 Some applications

1)  Medicine

- Inspection and interpretation of img obtained from X-rays, MRI or CAT scans

- Analysis of Cell img 

 

2) Agriculture

- Satellite/aerial views of land

- Inspection of fruit and vegetables

 

3) Industry

- Automatic inspection of items on a production line

- Inspection of paper samples

 

4) Law enforcement

- Fingerprint analysis

- Sharpening or De-blurring of speed-camera img

 

 

1.5 Aspects of image processing

1) Image Enhancement

: Processing img to be more suitable for a particular application (Look better)

- Sharpening or De-blurring an out of focus img

- Highlighting Edges

- Improving Contrast, Brightening img

- Removing Noise

 

2) Image Restoration

: Reversing the damage done to img by a known cause

- Removing of blur caused by linear motion

- Removal of optical distortions

- Removing periodic interference

 

3) Image Segmentation

: Subdividing img into constituent parts, Isolating certain aspect of img

- Finding lines, circles, particular shapes in img

- In an aerial photograph, Identifying cars, trees, buildings, or roads

 

 

1.6 An image processing task

1) Acquiring the image

: Producing a digital image from a paper envelope using a CCD camera, or a scanner, ...

 

2) Preprocessing

: Performing some basic tasks to render the resulting img more suitable for the job to follow

ex) enhancing the contrast, removing noise, identifying regions likely to contain the postcode

 

3) Segmentation

: Extracting postcode from img 

 

4) Representation and Description

: Extracting particular features to differentiate bw objects

ex) curves, holes, corners to distinguishing different digits which constitute a postcode

 

5) Recognition and Interpretation

: Assigning labels to objects based on descriptors & Assigning meanings to those labels

 

 

1.7 Types of digital images

1) Binary

- Each pixel is just 0(black) or 1(white)

- 1 bit per pixel → Very efficient to store

- Suitable for text (printed or handwrting), fingerprints, architectural plans

 

2) Greyscale

- Each pixel is a shade of grey, normally from 0(black) to 255(white)

- 8 bits (1 byte) per pixel → Very natural range for img file handling

- 256 different grey levels is sufficient for recognition of most natural objects

 

3) RGB (True color)

- Each pixel has a particular color, described by amount of RGB 0-255

- 255^3 = 16,777,216 different possible colors in img

- 24 bits (3 byte) per pixel

- Considered as consisting of a 'Stack' of three matrices

→ Every pixel has 3 values (RGB values)

 

4) Indexed

- Color map, or Color palette : A simple list of all the colors used in that img

- For convenience of storage and file handling

- Each pixel has a value which does not give its color, but an index to the color in the map

- It is convenient if img has 256 colors or less ( = only 1 byte index values to store )

- Without color map, img would be very dark and colorless

 

 

1.8 Image File Sizes

1) Image File Sizes tend to be large

 

2) The amount of information (# of bits) used in different image type of varying size 

- 512 x 512 binary image : 512 x 512 x 1 bits = 262144 bits = 32768 bytes = 32768 Kb ~ 0.033Mb

- 512 x 512 grayscale image : 512 x 512 x 1 bytes = 262144 bytes = 262.14 Kb ~ 0.262Mb

- 512 x 512 RGB color image : 512 x 512 x 3 bytes = 786432 bytes = 786.43 Kb ~ 0.786 Mb

 

3) A picture is worth 1000 words

A word contains 10 ASCII char (on average), and each char requires 8 bits of storage

→ Then, 1000 words contain 1000 x 10 x 8 = 8000 bits of information

≒ 283 x 283 binary img ≒ 100 x 100 greyscale img ≒ 58 x 58 RGB color img

 

1.9 Image Acquisition

How to get a picture 'into' a computer

1) CCD camera

Such a camera has, in place of the usual film, an array of photosites;

these are silicon electronic devices whose voltage output is proportional to the intensity of light falling on them

 

2) Flat bed scanner

Instead of the entire image being captured at once on a large array, a single row of photosites is moved across the image, capturing it row-by-row as it moves

 

 

1.10 Image Perception

1) Two basic steps of Image Perception

(1) Capturing the img with the eye

(2) Recognizing and Interpreting the img with the 'visual cortex' in the brain

 

2) Things to bear in mind

(1) Observed intensities vary as to background

- Grey scales differ from their surroundings (background)

- How much darker the square appears when it is surrounded by a light grey

- A single block of grey will appear darker if it were placed on a white background than black

 

(2) Observed Non-existent intensities as bars in continuously varying grey levels

- This img varies continuously from light to dark as we travel from left to right

- But, impossible for our eyes not to see a few horizontal edges in this image

 

(3) Our visual system tends to undershoot or overshoot around the boundary of regions of different intensities

Ex) A light grey blob on a dark grey background

As our eye travels from dark background to light region, boundary of region appears lighter than rest of it

↔ Going in the other direction, boundary of background appears darker than rest of it