Image Processing Reference
Highlight filter performs a different contrast increase for each component of each pixel in
the image. A component—let's take as an example the Red component—of the current pixel is
increased or decreased depending on the value of the Red component of the current pixel and
the two other filled with Red color pixels as in Figure 6 that shows a small 3 × 3 area within
an image. The pixels that are not filled with Red color in Figure 6 do not contribute to the new
value of the Red component of the current pixel.
FIGURE 6 Pixels (placed on the main diagonal) that count in computing the value of the con-
trast scale of each individual component (i.e., Red, Green, and Blue) of the current pixel.
Pixels that contribute to the new value of the current pixel in Figure 6 are placed in a diag-
onal manner. Because of this, both vertical and horizontal edges of the characters are detected.
For each component of each pixel, a contrast scale is computed separately. The way contrast
scale of each component of the current pixel ( x , y ) is computed is emphasized in Equations
x 1 y 1 , x 2 y 2 indicate the pixels with coordinates ( x − 1, y − 1) and ( x − 2, y − 2). For instance, r _ x 2 y 2
is the value of the Red component of the pixel with coordinates ( x − 2, y − 2).
left and top edges of the image. Once the contrast scale for each component of each individual
pixel in the image has been recorded, the algorithm is ready to be applied.
The starting point of Highlight filter algorithm is based on the fact that human eye is sens-
itive to a difference of at least 30 between the values of at least one of the same component of
two adjacent pixel when it comes to perceive and recognize characters. To be more specific, if
ifwe would have to write some characters on a background which is uniformly colored with the
intensity (R b ,G b ,B b ) = (0,0,0), the color of the text would have to be (R t ,G t ,B t ) = (0 + 40,31 + 0,0)