|
|
| Re: Read _ tiff with Sub_Rect Keyword [message #44879 is a reply to message #44876] |
Tue, 26 July 2005 05:06   |
raval.chintan
Messages: 54
Registered: May 2005
|
Member |
|
|
Hi Chris,
Thank You for this valuable suggestion, it is working fine, and is more
efficient. You can find the new code, for the opening the image, below:
pro read_geotiff
file = dialog_pickfile(filter=['*.tif', '*.tif'])
if file eq '' then return
print,systime(0)
res = query_tiff(file,info,geotiff=geostruct)
samples = info.dimensions[0]
lines = info.dimensions[1]
xper = 25
yper = 25
ximg = samples/xper
yimg = lines/yper
print,samples,lines,ximg,yimg
image = bytarr(3,ximg,yimg)
j=0
m=0
for i = 0, lines-1, yper+1 do begin
a = read_tiff(file,sub_rect=[0,i,samples,1])
k=0
for j =0 ,samples-1,xper+1 do begin
image[0,k,m] = a[0,j]
image[1,k,m] = a[1,j]
image[2,k,m] = a[2,j]
k++
endfor
m++
endfor
print,systime(0)
window,0,xsize=ximg,ysize=yimg
tv, image,/order, /true
end
Any new suggestions are welcome.
With Regards,
Chintan
Chris Torrence wrote:
> Hi Chintan,
>
> Well, you might try reading in complete scanlines (rows) instead of just
> doing tiles. For many TIFF files, the tiles are actually stored using
> scanlines. So for example your tilesize in the file might be 14000x1. So if
> you are trying to read in 500x500 chunks, it is extremely inefficient
> because it needs to read in 500 of the 14000x1 tiles, and then throw away
> most of the information. Then you go on to the next tile and it *again*
> reads in the same 500 14000x1 tiles.
>
> Also, even if your TIFF tilesize isn't 14000x1, it is sometimes still more
> efficient to read in entire scanlines because your operating system will
> tend to cache contiguous blocks of the file in memory.
>
> I think ENVI reads using scanlines.
>
> Hope this helps.
>
> -Chris
> Research Systems, Inc.
>
>
> <raval.chintan@gmail.com> wrote in message
> news:1122287818.648388.298350@g14g2000cwa.googlegroups.com.. .
>> Hi Ben,
>>
>> Thank You for your answer.
>>
>> Over here you are right , for reading the image file which has small
>> dimension, But my image dimension is 14000x14000 or more than that, and
>> i want to display this image in the size of 500X500 window. Here is my
>> code, try out with the image which has dimension more than 6000 X 6000.
>>
>> pro read_geotiff
>> file = dialog_pickfile(filter=['*.tif', '*.tif'])
>> if file eq '' then return
>> print,systime(0)
>> res = query_tiff(file,info,geotiff=geostruct)
>> samples = info.dimensions[0]
>> lines = info.dimensions[1]
>> xper = 25
>> yper = 25
>> ximg = samples/xper
>> yimg = lines/yper
>> print,samples,lines,ximg,yimg
>> image = bytarr(3,ximg,yimg)
>>
>> for i = 0, yimg*(yper-1), yimg do begin
>> for j = 0, ximg*(xper-1), ximg do begin
>> a = read_tiff(file,sub_rect=[j,i,ximg,yimg])
>> sx = j/xper
>> sy = i/yper
>> image[0:2,sx:sx+ximg/xper,sy:sy+yimg/yper] = a[0:2,$
>> 0:ximg-1:xper,0:yimg-1:yper]
>> endfor
>> endfor
>>
>> print,systime(0)
>> window,0,xsize=ximg,ysize=yimg
>> tv, image, /order, /true
>>
>> end
>>
>> If you know the other methode apart from this then please let me know.
>> Over here one solution is to make our program to read tiff file, but i
>> do not want to do this. Because if ENVI is using this read_tiff
>> function then there should be other method as I think ( Because envi is
>> taking less time to display the big tiff image).
>>
>> Regards
>> Chintan
>> Ben Tupper wrote:
>>> raval.chintan@gmail.com wrote:
>>>> Hi...
>>>>
>>>> I have a geo tiff file which containing the 14000 X 14000 Pixels
>>>> (Samples and Lines), I am reading it through the read_tiff function.
>>>> Over here i want to show that image on to the 500 X 500 window Means
>>>> by buffer for image will contain dimension [3,500,500]. For that I am
>>>> reading with the help of read_tiff function with the sub_rect keyword.
>>>> Where i m reading the pixels based on the ratio of 14000/500, but it is
>>>> taking to much time. While the same thing in ENVI it is taking less
>>>> time to read the image and display it. So is there any other method for
>>>> read that image fast. for that i have to write my own code in IDL to
>>>> read the tiff file?
>>>>
>>>> Regards,
>>>> Chintan Raval
>>>>
>>> Hello,
>>>
>>> I don't think I understand what it is that you are trying to do, but I
>>> would assume that IDL and ENVI are accessing the image using the same
>>> procedure. Here is an example of how to use the SUB_RECT keyword (which
>>> I use all the time on much smaller images with no problem.)
>>>
>>> file = FILE_SEARCH(!DIR, 'image.tif')
>>> whole = READ_TIFF(file[0])
>>> sub = READ_TIFF(file[0], sub_rect = [200, 200, 50, 100] )
>>> TV, whole
>>> TV, sub, 100, 0
>>>
>>> Is this example similar to how you are using the SUB_RECT keyword?
>>>
>>> Ben
>>
|
|
|
|
| Re: Read _ tiff with Sub_Rect Keyword [message #44888 is a reply to message #44879] |
Mon, 25 July 2005 08:32 |
Chris[2]
Messages: 39
Registered: August 2003
|
Member |
|
|
Hi Chintan,
Well, you might try reading in complete scanlines (rows) instead of just
doing tiles. For many TIFF files, the tiles are actually stored using
scanlines. So for example your tilesize in the file might be 14000x1. So if
you are trying to read in 500x500 chunks, it is extremely inefficient
because it needs to read in 500 of the 14000x1 tiles, and then throw away
most of the information. Then you go on to the next tile and it *again*
reads in the same 500 14000x1 tiles.
Also, even if your TIFF tilesize isn't 14000x1, it is sometimes still more
efficient to read in entire scanlines because your operating system will
tend to cache contiguous blocks of the file in memory.
I think ENVI reads using scanlines.
Hope this helps.
-Chris
Research Systems, Inc.
<raval.chintan@gmail.com> wrote in message
news:1122287818.648388.298350@g14g2000cwa.googlegroups.com.. .
> Hi Ben,
>
> Thank You for your answer.
>
> Over here you are right , for reading the image file which has small
> dimension, But my image dimension is 14000x14000 or more than that, and
> i want to display this image in the size of 500X500 window. Here is my
> code, try out with the image which has dimension more than 6000 X 6000.
>
> pro read_geotiff
> file = dialog_pickfile(filter=['*.tif', '*.tif'])
> if file eq '' then return
> print,systime(0)
> res = query_tiff(file,info,geotiff=geostruct)
> samples = info.dimensions[0]
> lines = info.dimensions[1]
> xper = 25
> yper = 25
> ximg = samples/xper
> yimg = lines/yper
> print,samples,lines,ximg,yimg
> image = bytarr(3,ximg,yimg)
>
> for i = 0, yimg*(yper-1), yimg do begin
> for j = 0, ximg*(xper-1), ximg do begin
> a = read_tiff(file,sub_rect=[j,i,ximg,yimg])
> sx = j/xper
> sy = i/yper
> image[0:2,sx:sx+ximg/xper,sy:sy+yimg/yper] = a[0:2,$
> 0:ximg-1:xper,0:yimg-1:yper]
> endfor
> endfor
>
> print,systime(0)
> window,0,xsize=ximg,ysize=yimg
> tv, image, /order, /true
>
> end
>
> If you know the other methode apart from this then please let me know.
> Over here one solution is to make our program to read tiff file, but i
> do not want to do this. Because if ENVI is using this read_tiff
> function then there should be other method as I think ( Because envi is
> taking less time to display the big tiff image).
>
> Regards
> Chintan
> Ben Tupper wrote:
>> raval.chintan@gmail.com wrote:
>>> Hi...
>>>
>>> I have a geo tiff file which containing the 14000 X 14000 Pixels
>>> (Samples and Lines), I am reading it through the read_tiff function.
>>> Over here i want to show that image on to the 500 X 500 window Means
>>> by buffer for image will contain dimension [3,500,500]. For that I am
>>> reading with the help of read_tiff function with the sub_rect keyword.
>>> Where i m reading the pixels based on the ratio of 14000/500, but it is
>>> taking to much time. While the same thing in ENVI it is taking less
>>> time to read the image and display it. So is there any other method for
>>> read that image fast. for that i have to write my own code in IDL to
>>> read the tiff file?
>>>
>>> Regards,
>>> Chintan Raval
>>>
>> Hello,
>>
>> I don't think I understand what it is that you are trying to do, but I
>> would assume that IDL and ENVI are accessing the image using the same
>> procedure. Here is an example of how to use the SUB_RECT keyword (which
>> I use all the time on much smaller images with no problem.)
>>
>> file = FILE_SEARCH(!DIR, 'image.tif')
>> whole = READ_TIFF(file[0])
>> sub = READ_TIFF(file[0], sub_rect = [200, 200, 50, 100] )
>> TV, whole
>> TV, sub, 100, 0
>>
>> Is this example similar to how you are using the SUB_RECT keyword?
>>
>> Ben
>
|
|
|
|
| Re: Read _ tiff with Sub_Rect Keyword [message #44892 is a reply to message #44888] |
Mon, 25 July 2005 03:36 |
raval.chintan
Messages: 54
Registered: May 2005
|
Member |
|
|
Hi Ben,
Thank You for your answer.
Over here you are right , for reading the image file which has small
dimension, But my image dimension is 14000x14000 or more than that, and
i want to display this image in the size of 500X500 window. Here is my
code, try out with the image which has dimension more than 6000 X 6000.
pro read_geotiff
file = dialog_pickfile(filter=['*.tif', '*.tif'])
if file eq '' then return
print,systime(0)
res = query_tiff(file,info,geotiff=geostruct)
samples = info.dimensions[0]
lines = info.dimensions[1]
xper = 25
yper = 25
ximg = samples/xper
yimg = lines/yper
print,samples,lines,ximg,yimg
image = bytarr(3,ximg,yimg)
for i = 0, yimg*(yper-1), yimg do begin
for j = 0, ximg*(xper-1), ximg do begin
a = read_tiff(file,sub_rect=[j,i,ximg,yimg])
sx = j/xper
sy = i/yper
image[0:2,sx:sx+ximg/xper,sy:sy+yimg/yper] = a[0:2,$
0:ximg-1:xper,0:yimg-1:yper]
endfor
endfor
print,systime(0)
window,0,xsize=ximg,ysize=yimg
tv, image, /order, /true
end
If you know the other methode apart from this then please let me know.
Over here one solution is to make our program to read tiff file, but i
do not want to do this. Because if ENVI is using this read_tiff
function then there should be other method as I think ( Because envi is
taking less time to display the big tiff image).
Regards
Chintan
Ben Tupper wrote:
> raval.chintan@gmail.com wrote:
>> Hi...
>>
>> I have a geo tiff file which containing the 14000 X 14000 Pixels
>> (Samples and Lines), I am reading it through the read_tiff function.
>> Over here i want to show that image on to the 500 X 500 window Means
>> by buffer for image will contain dimension [3,500,500]. For that I am
>> reading with the help of read_tiff function with the sub_rect keyword.
>> Where i m reading the pixels based on the ratio of 14000/500, but it is
>> taking to much time. While the same thing in ENVI it is taking less
>> time to read the image and display it. So is there any other method for
>> read that image fast. for that i have to write my own code in IDL to
>> read the tiff file?
>>
>> Regards,
>> Chintan Raval
>>
> Hello,
>
> I don't think I understand what it is that you are trying to do, but I
> would assume that IDL and ENVI are accessing the image using the same
> procedure. Here is an example of how to use the SUB_RECT keyword (which
> I use all the time on much smaller images with no problem.)
>
> file = FILE_SEARCH(!DIR, 'image.tif')
> whole = READ_TIFF(file[0])
> sub = READ_TIFF(file[0], sub_rect = [200, 200, 50, 100] )
> TV, whole
> TV, sub, 100, 0
>
> Is this example similar to how you are using the SUB_RECT keyword?
>
> Ben
|
|
|
|
| Re: Read _ tiff with Sub_Rect Keyword [message #44903 is a reply to message #44892] |
Fri, 22 July 2005 06:52 |
btt
Messages: 345
Registered: December 2000
|
Senior Member |
|
|
raval.chintan@gmail.com wrote:
> Hi...
>
> I have a geo tiff file which containing the 14000 X 14000 Pixels
> (Samples and Lines), I am reading it through the read_tiff function.
> Over here i want to show that image on to the 500 X 500 window Means
> by buffer for image will contain dimension [3,500,500]. For that I am
> reading with the help of read_tiff function with the sub_rect keyword.
> Where i m reading the pixels based on the ratio of 14000/500, but it is
> taking to much time. While the same thing in ENVI it is taking less
> time to read the image and display it. So is there any other method for
> read that image fast. for that i have to write my own code in IDL to
> read the tiff file?
>
> Regards,
> Chintan Raval
>
Hello,
I don't think I understand what it is that you are trying to do, but I
would assume that IDL and ENVI are accessing the image using the same
procedure. Here is an example of how to use the SUB_RECT keyword (which
I use all the time on much smaller images with no problem.)
file = FILE_SEARCH(!DIR, 'image.tif')
whole = READ_TIFF(file[0])
sub = READ_TIFF(file[0], sub_rect = [200, 200, 50, 100] )
TV, whole
TV, sub, 100, 0
Is this example similar to how you are using the SUB_RECT keyword?
Ben
|
|
|
|
| Re: Read _ tiff with Sub_Rect Keyword [message #44995 is a reply to message #44876] |
Wed, 27 July 2005 12:28 |
Chris[2]
Messages: 39
Registered: August 2003
|
Member |
|
|
Couldn't you eliminate the loop entirely using stride?
image[*, *, m] = a[*,0:samples-1:xper+1]
Or, even better, since you are inserting the "a" array into a contiguous
block of memory in "image", you can just use 0,0 as the subscripts:
image[0, 0, m] = a[*,0:samples-1:xper+1]
This will insert the entire "a" subarray into "image", starting at position
[0,0]. This is *much* faster than using index ranges, or even the *'s,
because IDL doesn't have to compute the locations internally. It just copies
the data as one block.
One other point. I think you want just "xper" in your loop, not "xper+1".
And finally, if your image isn't a multiple of "xper", your array indexing
will run off the end. So I think you really want to compute ximg and yimg
as:
ximg = (samples + xper - 1)/xper
yimg = (lines + yper - 1)/yper
-Chris
<meinel@aero.org> wrote in message
news:1122383746.592213.326040@g44g2000cwa.googlegroups.com.. .
> Isn't it more efficient to write
>
> for j =0 ,samples-1,xper+1 do begin
> image[0:2,k,m] = a[0:2,j]
> k++
> endfor
>
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
