Using a counting mask to calculate spacers is an efficient method. Where both accuracy and processing speed are concerned. A counting mask uses a 5×5 square matrix to capture patterns of spacers of a LCD plate [5]. There are three steps for the counting mask:

1. Marking on interesting patterns.

2. Expanding the mask for the spacers step by step.

Here we adopt the following method to perform the counting mask operation:

test(.) : to detect the pattern be spacer or background.

num : number of spacers

Yij : each pixel of ROI

testedge (.) : to detect the edges of the counting mask.

testedge(n2L→n5L) to detect the left edge of the counting mask.

testedge(nBL→nB5): to detect the bottom edge of the counting mask.

Nallowable: The allowable value in the counting mask.

L=1,B=5,R=5

repeat

for all Yij

if test ( Yij ) ==true

n13 = Yij

num=num+1

repeat

if testedge(n2L→n5L) ==true

L=L-1

untill testedge(n2L→n5L) ==false

repeat

if testedge(nBL→nB5)==true

B=B +1

untill testedge(nBL→nB5) ==false

repeat

if testedge(nR1→nRB)==true

R=R+1

untill testedge(nR1→nRB)==false

if (L++B+R)> Nallowable

error=error+1

until Yij out of ROI

The Counting mask is 5×5 square matrix (lack of upper right corner, as shown in Fig.12 which is used to detect patterns on the LCD plate. Every element on the mask can detect the performance of patterns and label it suitably, according to the sequence 1 to 5 as shown In Fig.13. If there is any spacers at the edge of the counting mask, the counting mask will expand bigger according to the sequence 6 to 8.

index