Vision Systems Flashcards
Why is feature extraction an important step in a classification task
Features have lower dimensions than the input images
Why do features need to be invariant to rotation, translation, and scaling
A feature could be located anywhere in image
Compactness can be calculated for what types of objects
binary valued shapes
what is the largest compactness value, and what shape is associated with it
1, circle
what does the hough transform do
finds objects inside the image with the same shape as its template
Which out of the HT and segmentation require a edge map for detection
HT
Simply, what does a low pass filter do
removes noise from images
how does a median filter clean noise
ignores extreme pixel values
Which type of filter use convolution
low pass filters
invarient moments can represent what objects
greyscale images and binary shapes
what is the aim of histogram equalisation
to flatten the histogram and improve the contrast and detail within the image
What are the 6 steps of involved in computer vision
sensing, pre-processing, segmentation, description, recognition, interpretation
describe the aim of pre-processing, what are the 2 main approaches
to clean up the image by improving detail and removing noise to allow data to be extracted
2 main approached = image enhancement and edge detection
what are the 3 types of edge
STEP - sudden change from light to dark
RAMP - gradual change from light to dark
ROOF - spike
What are area and perimeter invariant to
SCALE - no
TRANSLATION - ye
ROTATION - ye
