Data Translation, Inc.

pancakesnightmuteAI and Robotics

Nov 5, 2013 (3 years and 10 months ago)

76 views

Image Processing Concepts

Data Translation, Inc.

Basics of Image Processing

Image Processing Concepts

Types of Video Images


Monochrome


Image made up of varying shades of gray,
from black to white


Number of shades depends on resolution
of frame grabber

Image Processing Concepts

Monochrome Resolutions

Bits of
Resolution

Number of
Gray Levels

Black Value

White Value

1

2

0

1

4

16

0

15

6

64

0

63

8

256

0

255

12

4096

0

4095

16

65536

0

65535

Image Processing Concepts

Types of Video Images


RGB Color




Three sets of brightness signals


One for each: RED, GREEN, BLUE

Image Processing Concepts

Capturing Images


Terminology


Video formats


Frames (interlaced and non
-
interlaced)


Fields (odd and even)


Video signals

Image Processing Concepts

Standard Video Fundamentals



Light collides with the surface of an
image sensing device (camera)


Result: an electrical voltage level,
related to the amount of light hitting the
particular area of the surface, is
produced

Image Processing Concepts

Video Format



This information is then arranged into a
standard format and output from the
camera


Sync pulses are also added so that the
receiving device can recognize where
the sequence is in the frame data

Image Processing Concepts

Simplified Camera Diagram


Electron Beam

Video Signal

Target

Lens

Object

Image Processing Concepts

Pixels



Images are broken down into horizontal
lines


Lines are broken down into picture
elements, or pixels

Image Processing Concepts

Monochrome Pixels

Each pixel has a gray value. On 8
-
bit systems,
0=black and 255=white. All other values are
shades of gray.

Image Processing Concepts

Color Pixels

Each pixel contains 3 colored phosphors: RED, GREEN, and
BLUE. Each color receives a different intensity value (similar
to “gray scale” in monochrome image processing). The
resulting combinations determine which color we see.

Image Processing Concepts

Frames


Interlaced


Non
-
interlaced

Image Processing Concepts

Interlaced Image Frames


All odd
-
numbered lines are read from
top to bottom, followed by all even
-
numbered lines

Image Processing Concepts

Interlaced Image Frames


-

Diagonal lines are
active video

-

Horizontal lines are
blanking (beam off)

-

At the bottom of the
raster, the beam if off
and video begins its
vertical retrace
(vertical blanking)

Image Processing Concepts

Fields


Interlacing causes the frame to be
divided into two fields: odd and even


Each field is displayed sequentially
giving the perception that the frame is
updated twice as often as it really is

Image Processing Concepts

Fields



Each field updated every 1/60 or 1/50 s


Each frame updated every 1/30 or


1/25 s


60 (50) fields per second: 30 (25) odd
and 30 (25) even

Image Processing Concepts

Fields


This method reduces noticeable flicker
when displaying images


When working with graphics or thin
lines, flicker becomes extremely
noticeable

Image Processing Concepts

Fields



To reduce flicker:


Use horizontal lines that are wider than 1
pixel (2 lines??)


Use long
-
persistence monitor


Use non
-
interlaced monitor for graphics

Image Processing Concepts

Monitors



An electron beam scans the surface of
the display tube


A horizontal sync resets the beam to the
left
-
most side of the screen and then
moves it down to the next line


When a vertical sync is detected, the
beam is reset to the top, left
-
most point
of the screen

Image Processing Concepts

Diagram of Monitor Essentials

Phosphors

Video Signal

Gun

Grid

Electrons

Image Processing Concepts

Video Signals



A video signal contains a series of
analog TV lines


Lines are separated from one another
by a sync pulse called horizontal sync


Fields are separated by a longer sync
pulse called vertical sync

Image Processing Concepts

Typical Video Line

Active Pixel Region

Full Scan Area

Horiz.
Blanking

Blanking
Level

Horiz.
Sync

Image Processing Concepts

Video Signals


Digital video transfers several bits
(representing

pixel values)
simultaneously


Two voltage levels, Logic 0 and 1


Transmitted on individual TTL
(Transistor
-
Transistor Logic) lines or
pairs of lines in differential mode
(RS
-
422 standard, less noise)

Image Processing Concepts

Interfacing Input Devices with Frame
Grabbers


Video Formats:


RS
-
170 and CCIR


RS
-
170 RGB and CCIR RGB


NTSC and PAL

Image Processing Concepts

RS
-
170


Specifies all timing and voltage levels
for standard commercial video signals


Used as basis for most B&W video
equipment in the U.S.

Image Processing Concepts

RS
-
170


For 60 Hz television systems (North
American standard)


Frame consists of 525 lines and is
displayed once every 1/30 of a second


Each field contains 262.5 lines

Image Processing Concepts

RS
-
170


Each field also contains 9 sync lines (18
lines per frame) and 11 “no video” or
“blanking” lines


A video frame consists of 485 viewable
lines:


525


18 (sync)


22 (blanking) = 485

Image Processing Concepts

RS
-
170


For camera compatibility, most frame
grabber manufacturers design boards
which capture 480 lines


Therefore, lines are clipped at the top
and bottom of the image

Image Processing Concepts

RS
-
170 RGB


Three RS
-
170 type signals, one for
each of the additive primary colors


red, green, and blue


Red, green and blue images are
displayed simultaneously


Image manipulations must be
performed independently on all three
components

Image Processing Concepts

CCIR


CCIR


International Radio Consultative
Committee


50 Hz equivalent to RS
-
170


A frame consists of 625 lines


Subtracting sync and blanking lines
yields 544 lines of displayable video


Lines are clipped from top and bottom
to display 512 lines

Image Processing Concepts

NTSC


NTSC


National Television Standards
Committee


Standard specification for color signals


60 Hz


Single line input


Color is superimposed over the
monochrome (RS
-
170) signal

Image Processing Concepts

NTSC


Color can be removed by frame grabber
using chrominance filter


Three most popular NTSC uses:


Broadcast television


Cable television


VCRs

Image Processing Concepts

PAL (Phase Alternation Line)


50 Hz equivalent to NTSC


European standard

Image Processing Concepts

Summary of Standard Signals

Used
Where*

Color

Hertz

Lines In

RS
-
170

USA

No

60

1

RS
-
170
RGB

USA

Yes

60

3

CCIR

INTL

No

50

1

CCIR
RGB

INTL

Yes

50

3

NTSC

USA

Yes

60

1

PAL

INTL

Yes

50

1

*USA = US, Canada, Japan, Brazil INTL = Most other countries

Image Processing Concepts

Non
-
standard Video Signals


Input device tells frame grabber when to
digitize


Non
-
interlaced signal


All lines are read in succession to
create a frame


One type is referred to as “slow scan”

Image Processing Concepts

Non
-
standard Video Signals


The following control signals must be
provided by the user


Scan trigger


Clock enable


Pixel clock


Pixel value (analog)

Image Processing Concepts

Other Definitions

Image Processing Concepts

Picture Aspect Ratio


The relationship between the width and
height of a frame

4

3

Image Processing Concepts

Pixel Aspect Ratio


The relationship between the width and
height of a pixel

US

INTL

5:4

1:1

1:1

3:2

Image Processing Concepts

Chrominance Filter



Jumper
-
selectable circuit that removes
color information from NTSC signals.
Normally found on monochrome frame
grabbers to prevent interference with
the monochrome image


Implemented via a notch filter

Image Processing Concepts

Look
-
Up Tables (LUT’s)


Implements pixel (point) processing


One value goes in, another comes out

0 = 0

60 = 75

75 = 19

193 = 200

222 = 222

230 = 229

75

19

Image Processing Concepts

Input Look
-
Up Tables


Used for thresholding


Real
-
time processing


Add or multiply by a constant

Image Processing Concepts

Thresholding


A pixel operation used to reduce the
number of gray levels displayed


One example is binary thresholding,
resulting in either black or white

Image Processing Concepts

Binary Thresholding Example

0
-

160

0

161
-

255

255

LUT

Image Processing Concepts

Binary Thresholding Example #2

0
-

80

0

226
-

255

200

LUT

81
-

175

176
-

225

0

100

Image Processing Concepts

Contrast


Refers to the clarity (sharpness or
dullness) of an image


A result of the ratio of black to gray to
white

Image Processing Concepts

Histogram


Graphic representation of contrast

0 20 40 60 80 100 120 140 160 180 200 220 240 255

10

20

30

40

50

Number of pixels

0 = black,
255 = white

Image Processing Concepts

Histogram Equalization


Alters the histogram, thereby smoothing
the contrast

Image Processing Concepts

Zoom


Magnification of an image


Typical factors: 2, 4, or 8

Image Processing Concepts

Pan


Shifts image to left or right

Image Processing Concepts

Scroll


Shifts image up or down

Image Processing Concepts

Overlay



Graphics or text that can be added to an
image


Destructive and non
-
destructive

Image Processing Concepts

Area of Interest (AOI)


Also known as Region of Interest (ROI)
or Active Region of Interest


A portion of an image


Specific rows and columns form a
rectangular section to be worked on


Image Processing Concepts

Frame Buffer


An individual array of image data. Most
common are:


512 x 512 x 8 bits (256 Kb of memory)
used on older boards


640 x 480 x 8 bits (300 Kb of memory)
used on newer boards

Image Processing Concepts

Filtering


A method of massaging the image’s
data

Image Processing Concepts

Types of Filtering


Low pass


blur


High pass


sharpen


Laplacian


enhance all edges


Horizontal edge detection/enhancement


Vertical edge detection/enhancement

Image Processing Concepts

Group Processing


Works on a group of pixels at one time


Used for filtering

Image Processing Concepts

Kernel


Arithmetic grid used to perform filtering

1

0

4

4

6

6

5

9

9

-
1

-
1

-
1

-
1

9

-
1

-
1

-
1

-
1

16

Original Pixel
Values

Kernel

Resulting
Middle Pixel

Image Processing Concepts

Frame Averaging


Adds together several frames, then
divides by the number of frames. This
produces a less noisy image


True vs. Weighted

Image Processing Concepts

Logic Operations


Provide a pixel
-
by
-
pixel combination of
two images

Image Processing Concepts

Logic Operations

O = False

1 = True

AND

OR

XOR

0

1

0

0

0

1

0

1

0

1

0

0

1

1

1

1

0

1

0

0

1

1

1

0

Image Processing Concepts

Hue Saturation Intensity (HSI)


“Human view of colors”


Rather than specifying a color as
percentages of red, green and blue,
they are specified as “dark magenta” or
“light aqua”


Takes the same number of bits to store
an HSI image as an RGB one

Image Processing Concepts

Conclusion

Additional image processing questions?
Contact Data Translation at

(800) 525
-
8528