how to describe matrix values as if signle value ?

[vhelp]

Let me explain the convoluting subject title..

I'm having a problem in the theory of mathematics, but don't know exactly the defination or terms to describe the problem(s) so I think it best if I try and explain it in detail (below) and maybe those who are gurus of math/equations/etc., can realize what I'm asking for and advise.

I coded up an algorithm that performs a single pixel 'compare' function. It is based on a range value (thresh: 0-255) and if within 'thresh' range, it considers the value true, as in (if A=B then True) ..

WHERE:

pA=15, and pB=20, and vThresh=5, and function pRange(pA, pB, vThresh): boolean

If pRange(pA, pB, vThresh) then doRoutineToPixel(image1)

So lets say that pB could be anything like: 15, 16, 17, 18, 19, or 20, then that is the 'within range' I want to consider, true, and then perform an action against the pixel in question.

Now, I've been using the routine above (among many others I coded up) for quite some time in a few of my image processing projects. But now, I want to extend it as a matrix type 'compare' function in other applications as well because I know (the idea/theory) can be enhanced to perform even better results.

The matrix I am considering is a 2x2 and 3x3 group of pixels. And these group of pixels (matrix) would be treated as if a single pixel. That's the theory that I'm having trouble grasping and making into a function. And unfortunately, when it comes to math, I am very slow.

I realize that this could be considered a motion-detection type routine, but that is not exactly what I'm looking at in this discussion. I am also interested in MD, but not at this junction of this endeavor.

. . .

When I am debuggin certain routines, I lean on my MS Excel as the sculpturing tool because of its great flexability and ease of building ones ideas/ methods/ algorithems, etc with. But, for this new or enhanced idea I am having trouble building the theory through excel, though many of my ideas are first tested/built inside an excel sheet. But I think my brain is finally overloaded. Basically, I'm stuck.

. . .

So, what I have so far, (using excel) is a simple 2x2 grid, spread across temporaly, and each group (matrix) are summed Horizontally and Virtically. And when both angles are added, total to the value of 8, seen in these crude example layouts, below.



* in this example, I would assume that all the images groups are equal, that is, they are the same, because they all have the same value of '8' and so nothing would be needed done on 'em.



* in this example, each of the matrix's are different. I would catagorize this as a 'big difference' or out-of-range.



* in this example, even though the values are different, they are only minor. And the last two matrix's, even though they are equal in summation, they placement of the different values are different.

The values you see in the example above is just to simplify due to char space limitation, etc. In images, we all know how diverse and scrabled they truely are. Especially when in noisy analog sources. But my concerns are with the theory and getting the concept understood so that I can move forward in this project.

I'm thinking that the 8 value is what I would use to identify that group of pixels (matrix) as if a signle pixel, but I'm sure. I think there is more to it than what I'm seeing at the momnet and it may not apply similary as in the 'single-pixel' analigy, and perform the same as my pRange() routine, above.

Any and all suggestion / adice would be greately appreciated, thanks in advance,

-vhelp 5134

[minidv2dvd]

i don't get the 2x2 graph, it appears to be 3x3 to me? are the pB's in []'s? and if it's only 2x2 wouldn't the sum of all 4 pB results be a better total to use?

[vhelp]

thank you for catching that, I'm sorry I messed up on the graph. Its hard to align things in this forums boards text limitations. let me try and fix it up a bit. I had some additional notes I was composing while you were reading.

-vhelp 5135