A Binary Image is one whose pixels take on a value of 0 or 1 (or some other 2 valued set of numbers....like 0 and 255).

Why Binary Images?

1. Analysis: Use to separate objects from the background. E.g. OCR, Object Recognition by Silhouette Tracing.
2. Art.
3. Reduced Storage and faster calculations

How to Create a Binary Image Binary Image = Thresholded Original Image

Thresholding an Image binary(r,c) = binary image pixel (r,c) f(r,c) = original image pixel (r,c) T = threshold value at every pixel location (r,c) in the image set: binary(r,c) =

Example Save have a greyscale image with 256 greylevels i.e. p(r,c) = {0, 1, 2, .... 255} say T = 100 then if p(r,c) <= 100 -----> binary(r,c)=1 else if p(r,c) > 100 -----> binary(r,c)=0

How to pick T? Want T such that only pixels belonging to object you want to highlight are greater than T. Trial and Error Can use the Image's Histogram to Pick T.

Pick T Threshold using Histogram Look at a plot of h(m) and look for a valley to separate the objects pixels from the rest of the image. For example, suppose h(m) takes on the following bi-modal shape. What about a Color Image? Color image: f(r,c) = (red,green,blue) e.g. f(r,c) = (255,255,255)= white color       f(r,c) = (0, 0, 0) = black color 1) Create a histogram for each color and create three binary images one for each color field: h red(), h green(), h blue() 2) Convert color to greyscale first as shown below and then threshold the new greyscale image:   greyscale value = f'(r,c) = (red + green + blue)/ Example: Find the coin Histogram and selected threshold level, Pick Threshold right where it seperates lighter backbround from darker coin pixels.