Red Eye

1
Session 15

• Red Eye
• (working with conditionals)

2
Remove Red Eye

• Red eye is when the flash from the camera is
  reflected from the subjects eyes
• We want to change the red color in the eyes to
  another color

3
Red Eye Algorithm

• Use the explorer to find the RGB values of the
  red eyes
• For every pixel in the Picture find the distance
  between the current color and our definition of
  red
• And change the color of the current pixel only if
  the current color is within some distance to the
  desired color
• Wait, we need to know TWO new things in order to
  do this!

4
New thing 1 Conditional Execution

• Sometimes we want a statement or block of
  statements executed only if some expression is
  true
• We can use the if statement in Python
• if (colorDistance lt value)
• Statement 1
• Statement 2, etc
• next statement

false
if (expression)
true
Statement or block
statement
5
New thing 2 Color Distance

• The distance between two points is computed as
• Square root of (( x1 x2)2 (y1 y2)2)
• The distance between two colors can be computed
• Square root of ((red1 red2)2 (green1-green2)2
  (blue1 blue2)2)
• There is a method in the Pixel class to do this
• dist distance(color1,color2)

6
removeRedEye() method

• def removeRedEye(picture)
• redEye makeColor(155,25,35)
• for pix in getPixels(picture)
• if (distance(redEye,getColor(pix)) lt 50)
• setColor(pix,black)

7
Oops, MOSTLY got rid of the red eye but also
8
Detailed Red Eye Algorithm

• Loop over a subset of the pixels
• Requires we give it starting and ending points
  for both x and y
• This allows us to leave the shirt alone while
  also using a wider distance for redEye

9
Remove Red Eye Method

• def specificRemoveRedEye(pic,startX,startY,endX,en
  dY)
• redEye makeColor(155,25,35)
• for x in range(startX,endX1)
• for y in range(startY,endY1)
• pixel getPixel(pict,x,y)
• if (distance(redEye,getColor(pixel)) lt
  75)
• setColor(pixel,black)

10
How would a computer see
One way computers see is to attempt to
determine the boundaries between items in an
image.
11
Edge Detection

• Loop through all the pixels in the picture
• Calculate the average color for the current pixel
  and the pixel at the same x but y1.
• Get the distance between the two averages
• If the distance is greater than some value turn
  the current pixel black
• Otherwise turn the current pixel white

12
Conditionals

• What is different about the conditional statement
  in this problem versus the conditional statement
  in the removeRedEye() problem?
• This has two situations

13
Use if and else for two possibilities

• Sometimes you want to do one thing if the
  expression is true
• and a different thing if it is false

false
if (expression)
else
true
Statement or block
Statement or block
statement
14
Edge Detection Algorithm

• To find areas of high contrast
• Try to loop from row 1 to row height
• Loop from x 1 to x width 1
• Get the pixel at the x and y (top pixel)
• Get the pixel at the x and (y 1) bottom pixel
• Get the average of the top pixel color values
• Get the average of the bottom pixel color values
• If the absolute value of the difference between
  the averages is over a passed limit
• Turn the pixel black
• Otherwise turn the pixel white

15
Edge Detection Exercise

• Write a method edgeDetection that takes an input
  limit
• And turns all pixels black where the absolute
  value of the difference between that pixel and
  the below pixel is greater than the passed limit
• And turns all pixels white where the absolute
  value of the difference between that pixel and
  the below pixel is less than or equal the passed
  limit
• Pixel has a getAverage() method that returns the
  average of the three colors at the pixel

16
Edge detection (an alternative to Program 40)

• def topDownEdge(source,limit)
• for y in range(1,getHeight(source))
• for x in range(1,getWidth(source)1)
• top getPixel(source,x,y)
• bottom getPixel(source,x,y1)
• if (distance(getColor(top),getColor(bottom
  )) lt limit)
• setColor(top,white)
• else
• setColor(top,black)

17
What about the other way?

• Create another method for simple edge detection
• This time compare the current pixel with the one
  to the right (x1)
• How do you need to change the nested loop?
• Do you get a different result?

18
Comparing left/right instead of up/down

• def leftRightEdge(source,limit)
• for x in range(1,getWidth(source))
• for y in range(1,getHeight(source)1)
• left getPixel(source,x,y)
• right getPixel(source,x1,y)
• if (distance(getColor(left),getColor(right
  )) lt limit)
• setColor(left,white)
• else
• setColor(left,black)