Rasterexpressionevaluator expression examples

mark2atsafe wrote:

I'm currently helping out with updates to the transformer help content. I've suggested that we do raster next and I will be sure to get some examples in the RasterExpressionEvaluator. Like you I just had to scramble around to find the right syntax to use, so I know how important it is to include clear examples.

So keep an eye out for changes in the not-too-distant future.

Here's some content I put together and will pass on to the documentation team for inclusion into the help file. Until I investigated I didn't realize that you could use attributes in an expression, or that || means an OR statement!

 

 

RasterExpressionEvaluator Info + Examples

 

 

|| means or

 

&& means and

 

== means equal

 

!= means not equal

 

< means less than

 

> means greater than

 

 

A[x] means band x of raster feature A

 

B[x] means band x of raster feature B

 

 

Bands start counting at zero (so a three-band raster is A[0], A[1], and A[2])

 

 

A simple statement might be like:

 

 

255

 

(set the value of the current cell to be 255)

 

 

A[0]+1

 

(set the value of the current cell to be Raster A, band zero, plus 1)

 

 

"if" statements allow a value to be set depending on conditions

 

if (conditions, pass value, fail value)

 

 

if (A[0] < 255, 200, 255)

 

(if Raster A, band zero is less than 255, then set the current cell to 200, else leave it as 255)

 

 

if (A[0]<50 || A[1]<50 || A[2]<50, 0, 255)

 

(if Raster A, band zero is less than 50 OR band one is less than 50 OR band two is less than 50, then set the current cell to 0, else set it to 255)

 

 

Bands in two raster features can be referenced in the statement

 

 

if (A[0]!=B[0] && A[0]<255, 0, A[0])

 

(if Raster A band zero is not equal to Raster B band zero, AND Raster A band zero is less than 255, then set the current cell to 0, else set it to the value of Raster A band zero)

 

 

Standard maths operators are permitted:

 

if (A[0] - B[0] == 255, 0, A[0]/5)

 

(if Raster A band zero minus Raster B band zero equals 255, then set the current cell to 0, else set it to the value of Raster A band zero divided by five)

 

 

Attributes can be referenced using @Value and either A: or B: to refer to the attribute name

 

A[0]*@Value(A:myattr)

 

(multiply Raster A band zero by the attribute myattr on raster A)

 

 

 

Practical Examples

 

255-A[0]

 

(Inverts an 8-bit band. For a 24-bit RGB image, repeat for A[1] and A[2] to invert the full image, repeat for A[3] too if it is a 32-bit RGBA image)

 

 

if ((A[0]+A[1]+A[2])==0, 255, A[0])

 

(Convert black values into white in a 24-bit RGB image. Repeat for A[1] and A[2] to invert the full image)

 

 

A[0]*1.5

 

(Increases the brightness of the red band in an RGB image. Repeat for A[1] and A[2] to brighten the full image)

 

 

@sqrt((A[0]*A[0]) + (B[0]*B[0]))

 

(Combines two raster cells together. An alternative to average the values would be (A[0]+B[0])/2)

 

 

if(A[0]==0,0,A[0]/@Value(A:_height+1))

 

(Classify raster DEM cells into groups where group size is defined by the _height attribute)

 

mark2atsafe wrote:

Here's some content I put together and will pass on to the documentation team for inclusion into the help file. Until I investigated I didn't realize that you could use attributes in an expression, or that || means an OR statement!

 

 

RasterExpressionEvaluator Info + Examples

 

 

|| means or

 

&& means and

 

== means equal

 

!= means not equal

 

< means less than

 

> means greater than

 

 

A[x] means band x of raster feature A

 

B[x] means band x of raster feature B

 

 

Bands start counting at zero (so a three-band raster is A[0], A[1], and A[2])

 

 

A simple statement might be like:

 

 

255

 

(set the value of the current cell to be 255)

 

 

A[0]+1

 

(set the value of the current cell to be Raster A, band zero, plus 1)

 

 

"if" statements allow a value to be set depending on conditions

 

if (conditions, pass value, fail value)

 

 

if (A[0] < 255, 200, 255)

 

(if Raster A, band zero is less than 255, then set the current cell to 200, else leave it as 255)

 

 

if (A[0]<50 || A[1]<50 || A[2]<50, 0, 255)

 

(if Raster A, band zero is less than 50 OR band one is less than 50 OR band two is less than 50, then set the current cell to 0, else set it to 255)

 

 

Bands in two raster features can be referenced in the statement

 

 

if (A[0]!=B[0] && A[0]<255, 0, A[0])

 

(if Raster A band zero is not equal to Raster B band zero, AND Raster A band zero is less than 255, then set the current cell to 0, else set it to the value of Raster A band zero)

 

 

Standard maths operators are permitted:

 

if (A[0] - B[0] == 255, 0, A[0]/5)

 

(if Raster A band zero minus Raster B band zero equals 255, then set the current cell to 0, else set it to the value of Raster A band zero divided by five)

 

 

Attributes can be referenced using @Value and either A: or B: to refer to the attribute name

 

A[0]*@Value(A:myattr)

 

(multiply Raster A band zero by the attribute myattr on raster A)

 

 

 

Practical Examples

 

255-A[0]

 

(Inverts an 8-bit band. For a 24-bit RGB image, repeat for A[1] and A[2] to invert the full image, repeat for A[3] too if it is a 32-bit RGBA image)

 

 

if ((A[0]+A[1]+A[2])==0, 255, A[0])

 

(Convert black values into white in a 24-bit RGB image. Repeat for A[1] and A[2] to invert the full image)

 

 

A[0]*1.5

 

(Increases the brightness of the red band in an RGB image. Repeat for A[1] and A[2] to brighten the full image)

 

 

@sqrt((A[0]*A[0]) + (B[0]*B[0]))

 

(Combines two raster cells together. An alternative to average the values would be (A[0]+B[0])/2)

 

 

if(A[0]==0,0,A[0]/@Value(A:_height+1))

 

(Classify raster DEM cells into groups where group size is defined by the _height attribute)

 

Hi Mark, this is really useful thank you.

 

 

Is it possible to use two expressions with different outcomes? In this example, I want to set red pixels to 255,255,255, blue pixels to 50,50,50, and any other pixels to 0,0,0.

 

 

What I'm roughly after is this:

 

 

if (A[0]>250 && A[1]==0 && A[2]==0,255,0) || if (A[0]==0 && A[2]==255),50,0)

 

 

ld wrote:

Hi Mark, this is really useful thank you.

 

 

Is it possible to use two expressions with different outcomes? In this example, I want to set red pixels to 255,255,255, blue pixels to 50,50,50, and any other pixels to 0,0,0.

 

 

What I'm roughly after is this:

 

 

if (A[0]>250 && A[1]==0 && A[2]==0,255,0) || if (A[0]==0 && A[2]==255),50,0)

 

 

Not like that I don't think . But what you may be able to do is something like:

 

 

if (A[0]>250 && A[1]==0 && A[2]==0,255,(if (A[0]==0 && A[2]==255),50,0))

 

 

Haven't tried it but I think it's worth a go. I'll try it later too when I get a moment.

 

mark2atsafe wrote:

Not like that I don't think . But what you may be able to do is something like:

 

 

if (A[0]>250 && A[1]==0 && A[2]==0,255,(if (A[0]==0 && A[2]==255),50,0))

 

 

Haven't tried it but I think it's worth a go. I'll try it later too when I get a moment.

 

Thanks Mark, but this didn't work unfortunately. I imagine it must be possible however so I would welcome any other suggestions.

 

 

Currently I'm just creating an average of all bands rather than a specific colour, which fulfils the requirement.

 

 

ld wrote:

Thanks Mark, but this didn't work unfortunately. I imagine it must be possible however so I would welcome any other suggestions.

 

 

Currently I'm just creating an average of all bands rather than a specific colour, which fulfils the requirement.

 

 

Actually, I think that does work. At least, I simplified it somewhat and can confirm that the idea of putting an extra "if" statement in place of a new value works.

 

 

Here I used 100 instead of 250 (it fit my test data better) and left out the && parts:

 

 

 

...and that worked just fine:

 

 

 

Here the light parts are 255,255,255; the dark parts are 0,0,0; the grey parts are 55,55,55. So, yeah I left out the && parts, and I don't think it's quite the same as what you were asking for, but I don't see that would stop the overall method from working (but I will try some more to check).

 

 

ld wrote:

Thanks Mark, but this didn't work unfortunately. I imagine it must be possible however so I would welcome any other suggestions.

 

 

Currently I'm just creating an average of all bands rather than a specific colour, which fulfils the requirement.

 

 

Hey @ld - yes, I think this is what you need:

 

 

 

The only difference is I used different limits (Red: >150, <100, <100 and Blue: <100, < 100, >100) because I had difficulties finding an image with pure red or pure blue in it. It's surprising how a cell can look red, but query it and the value is 125,34,42 - nowhere near the limit I would expect. Oh and I set the blue result to 125 to make it more visible for me.

 

 

Anyway, those expressions will do what you need, although you might need to tweak them a bit. Hope this helps.

 

mark2atsafe wrote:

Hey @ld - yes, I think this is what you need:

 

 

 

The only difference is I used different limits (Red: >150, <100, <100 and Blue: <100, < 100, >100) because I had difficulties finding an image with pure red or pure blue in it. It's surprising how a cell can look red, but query it and the value is 125,34,42 - nowhere near the limit I would expect. Oh and I set the blue result to 125 to make it more visible for me.

 

 

Anyway, those expressions will do what you need, although you might need to tweak them a bit. Hope this helps.

 

Thanks Mark, works perfectly