Monthly Archives: September 2018

The Project – Initial thoughts

Posted on by

Our project is describes as follows:

‘Find the left eye in the first image and track it through the image sequence’

The image sequence described contains a person talking about cats (I think) but the solution required needs to be generally applicable.

My initial thoughts are two create a program that can isolate a unique colour in conjunction with recognising patterns within the detail. of the image but I am sure there are many potential approaches that can be taken. So…

 

 

 

 

 

 

 

After research more specifics will follow…

Week 2 -Lab Material

Posted on by
Reply

This week we recapped on the theory we we learned last week and got some practical experience with OpenCV.

We got to explore various colour spaces along with some basic manipulation of images that we loaded into and exported from the system, as was reflected within the lab tasks.

Task 1

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
# 1. Read in any image
 
# Also saves read image as "imagecopy"
 
import numpy as np
import cv2
from matplotlib import pyplot as plt
from matplotlib import image as image
import easygui
 
I = cv2.imread("colors.jpg")
 
cv2.imwrite("imagecopy.jpg",I)
 
cv2.imshow("image", I)
key = cv2.waitKey(0)

 

Task 2

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
# 2. Convert the image to YUV
 
import numpy as np
import cv2
from matplotlib import pyplot as plt
from matplotlib import image as image
import easygui
 
I = cv2.imread("colors.jpg")
YUV = cv2.cvtColor(I, cv2.COLOR_BGR2YUV)
 
cv2.imwrite("imageYUV.jpg",I)
 
cv2.imshow("cvtColor", I)
key = cv2.waitKey(0)

 

Task 3

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
# 3. Extract the y Channel (Luminance)
 
import numpy as np
import cv2
from matplotlib import pyplot as plt
from matplotlib import image as image
import easygui
 
I = cv2.imread("colors.jpg")
YUV = cv2.cvtColor(I, cv2.COLOR_BGR2YUV)
 
cv2.imshow("Y", YUV[:, :, 0])
# cv2.imshow("U", YUV[:, :, 1])
# cv2.imshow("V", YUV[:, :, 2])
key = cv2.waitKey(0)
 
# Alternative method
# y,u,v = cv2.split(YUV)
# cv2.imshow("Y", y)
# cv2.imshow("U", u)
# cv2.imshow("V", v)
# key = cv2.waitKey(0)

 

Task 4

1
2
3
4
5
6
7
8
9
10
11
12
13
# 4. Convert the original image to grayscale (intensity)
 
import numpy as np
import cv2
from matplotlib import pyplot as plt
from matplotlib import image as image
import easygui
 
I = cv2.imread("colors.jpg")
J = cv2.cvtColor(I, cv2.COLOR_BGR2GRAY)
 
cv2.imshow("image", J)
key = cv2.waitKey(0)

 

Task 5

1
2
3
4
5
6
7
8
9
10
11
12
13
14
# 5. Show the two on the screen using OpenCV's imshow function
 
import numpy as np
import cv2
from matplotlib import pyplot as plt
from matplotlib import image as image
import easygui
 
I = cv2.imread("colors.jpg")
J = cv2.cvtColor(I, cv2.COLOR_BGR2GRAY)
 
cv2.imshow("image1", J)
cv2.imshow("image2", I)
key = cv2.waitKey(0)

 

Cones

Posted on by
Reply

The lecturer mentioned in class this week that Women were able to see a wider variety of colour than men.

This reminded me of a workshop I attended organised by Screen Skills Ireland a few years back which spoke about the difference between the colour space of Standard Definition and High Definition TV.

The facilitator started his initial lecture by talking about the anatomy of the human eye and how we perceive colour through a collection of cones and rods within the retina.  As it turns out Women generally have more cones which means expanded sensitivity within the red area of the colour spectrum.

Week 1 – My old friend BGR

Posted on by
Reply

This week we were introduced to the program which we are going to be using throughout the module OpenCV. While using this program we were advised to be aware that when operating within the more familiar RGB colour space this program preferred to mix things up a little for reasons best known to itself and operate within a BGR colour space. Which offers no significant problem until you decide to output into a program which uses RGB and then some kind of conversion process is required.

This reminded me of my youth as an intern working briefly at the BBC. I noticed one day that they were also very fond of this particular arrangement of these three components, which I would usually notice after I had hurriedly wired video monitoring equipment incorrectly, much to the annoyance of the chief engineer.

I asked him one day why they preferred using a BGR labelling scheme when everybody else seemed to prefer using RGB. He looked at me quizzically, slowly put down his cup of tea, took a long drag from his cigarette and replied “Because we’re the f**king BBC, that’s why’.

Welcome to my image processing blog!

Posted on by
Reply

I come from a Broadcast TV/Video background, and I am looking forward to exploring a field which I would have some familiarity with from a new perspective.

When I began my TV career, the idea of digital imagery would be regarded as something quite exotic and unusual, at that time we would be more familiar with it’s fuzzier counterpart the analogue image, which originated from celluloid or was scanned with an electron beam, that would then be encoded and fired down the wire with the aid of vector based technology to wherever it needed to go.