Max Diference = Mininum Value Central

gray

50	255	10
66	75	30
80	44	1

Sqrt(50-75)

Max Diference = Mininum Value – Central
Max Dif = 75-1

75-50 = 25*LumFactor

Possible Structures. Assume smaller then central = 0, bigger then central = 1

50 is smaller then 75, therefore = 0	255 is bigger then 75, therefore = 1	10
66	75	30
80	44	1

Turns onto

Convert it to binary we have

Max Value = 495

Or we can calculate the standard deviation of the difference instead

Equals to

Calculate the std of all the differences except the central number 75

Divide the central number with the standard dev
σ: 75.254567967666
DivisionRatio = 75.254567967666/75 = 1.003394

Now we divide each value with the standard deviation (except the central value 75)

Etc

2.38379 - 0.066216 = 2.317574

Value1 = (0.331082-0.066216)/(2.317574) * 255 = 29.14286
Value2 = (2.38379-0.066216)/(2.317574) * 255 = 255
Value3 = (0.860813-0.066216)/(2.317574) * 255 = 87.42857
Value4 = (0.066216-0.066216)/(2.317574) * 255 = 0

Ok, now we have a whole new set of locally structure data biased on the centered one (Whose value is still 75)

Passing onto our table we wil have:

29.14286	255	87.42857
....	75	....
0	...	....

And again, we can do the smaller then 75 thing to make it binary

Turns onto

Possible Structures. Assume smaller then central = 0, bigger then central = 1

And we compute he the true binary of it.

Ex: Say we have a binary value resulting in 48

This is our texture/colormatch data. All we have to do is create a table (matrix 9x9) that carries the Luma, Hue and Saturation (or Lab) values and associate it with it the value 48.

Whatever pixel whose color value = 75 and Binary Difference Value = 48 we then search on our tables to be filled.

So we can have a maximum of 256 binary difference values from each pixel on a total of at least 256*256 distinct data with similar structure from each other

Can we do more ? Yes we can.

We can take the Difference table to build a 81x81 pixels, separated with 9x9 quadrants and compare the diff result on each other

Math =

X1	X2	X3
Y1	k	Y2
Z1	Z2	Z3

Mean = ((X1-k)+(X2-k)+(X3-k)+(Y1-k)+(Y3-k)+(Z1-k)+(Z2-k)+(Z3-k))/8

1/8 (1/8 (8 k - X1 - X2 - X3 - Y1 - Y3 - Z1 - Z2 - Z3) + X2)^2

F = (8*k-(SumPixels))/8
8*F = 8*k-SumPixels
SumPixels = 8*k-8*F
Variance = (((F+X1)^2)/8) + (((F+X2)^2)/8) +
(((F+X1)^2)/8)+(((F+X2)^2)/8)+(((F+X3)^2)/8)+(((F+Y1)^2)/8)+(((F+Y3)^2)/8)+(((F+Z1)^2)/8)+(((F+Z2)^2)/8)+(((F+Z3)^2)/8)

k^2 - S^2/64 + 1/8 (X1^2 + X2^2 + X3^2 + Y1^2 + Y3^2 + Z1^2 + Z2^2 + Z3^2)

Outro filtro

T1 = (X1^2)/8+(X2^2)/8+(X3^2)/8+(Y1^2)/8+(Y3^2)/8+(Z1^2)/8+(Z3^2)/8+(Z3^2)/8