Please enable JavaScript.
Coggle requires JavaScript to display documents.
Computer Science Exam Preparation Computer Vision, Extra Resource Computer…
Computer Science Exam Preparation Computer Vision
Thresholding
Pixels
"segmentation"
Assigning pixels
numbers
Black and white
if > threshold (over)
assigns white
If < threshold (under)
assigns black
(not always)
values
be depicted from
0 to 255
0 = Black
255 = White
RGB Values
Lights, camera
Human Eyes
senses
RGB (Red, Green, Blue) light
special cells
perception
Light and dark colors
imperfect
overlapping view
Digital Cameras
silicon chip
Detects RGB
imperfect
white light
look like blobs
loses detail
Conclusion
both imperfect
improving
research
Big Picture
growing range
used in fields
e.g. healthcare, security
"see"
translate languages
imperfect
how: pixels (images close to letters)
thresholding
black and white pixels
face recognition
steps
identification
greyscale
crop
mapped
identity setted up
examples
google maps
security cameras
google maps:
blurs faces
Security camera
identification
Edge Detection
recognition
Edges around objects
or text
Depth
measures
distance
between items & camera
Summary
improve
cameras
resolution
low light
infra-red
< noise
cheaper
old ideas
relevant
e.g. "k-means"
1967 :checkered_flag:
1975 digicam
introduction
input
past
keyboard
now
camera
Noise
challenge
pixels
black and white
appearance
becomes darker or lighter
caused by
interference
"grainy" effect
some devices
sensitive to light
(high ASA/ISO setting)
unwanted pixels
not recognisable
Fixable
predictions
good and bad locations pixels
combines RGB light
into a "grayscale" image
saves time
Other techniques
predicts values
of neighboring pixels
Filters
Mean
problem
bigger square
causes blurring
effects edges/details
predicts
nearby pixels
are alike other pixels
Median
problem
not given values
ignores light or dark value
organises values
chooses dark or light value
Gaussian
predicts
less away pixels
similar
far away pixels
less similar
mathematically balanced
Extra Resource Computer Vision
Three steps
aquire
process
understand
Other Tools
LIDAR
uses
light frequencies
Infrared
used for heat
Radar
uses
radio frequency
narrow range
long distances
like weather
Ultraviolet
measures
small distances
Text recognition
Key aspects