FRIHOST • FORUMS • FAQ • TOS • BLOGS • DIRECTORY
You are invited to Log in or Register a Frihost Account!

Help in image processing (Expert only please)

   Frihost Forum Index -> Scripting -> Others
 


bukaida
Please help me with the following problems--

1. Does an image viewer decompresses the compressed image(like jpeg) before displaying it? or is it possible to extract the image information in the compressed form itself?

2. Please help me with the code (in any web-based language) which can identify the textual information in a scanned image using jpeg2000 or any other wavelet based compression technique.

3. Is it possible to extrat a subarray from the compressed image itself without decompressing it? Please help.
Indi
bukaida wrote:
Please help me with the following problems--

1. Does an image viewer decompresses the compressed image(like jpeg) before displaying it? or is it possible to extract the image information in the compressed form itself?

It must always be decompressed. Sometimes that's done in hardware, but it must always be done.

bukaida wrote:
2. Please help me with the code (in any web-based language) which can identify the textual information in a scanned image using jpeg2000 or any other wavelet based compression technique.

That requires OCR, optical character recognition software. That is very complicated, using edge detection and pattern recognition algorithms. It is unrealistic to expect such a thing to be written in any web-based language.

bukaida wrote:
3. Is it possible to extrat a subarray from the compressed image itself without decompressing it? Please help.

For JPEG2000 yes, but i believe there are restrictions (i haven't written a JPEG2000 encoder/decoder personally). Remember that there are varying levels of compression. It may not be possible to get exactly pixels 131 to 137, but you will be able to get (for example) pixels 130 to 140 or 100 to 150 depending on how the image was compressed. So yes, you can extract a sub-array, but you may need to extract a little more than what you exactly wanted.
bukaida
Quote:
It must always be decompressed. Sometimes that's done in hardware, but it must always be done.

If that be the case, then where does it store the decompressed picture?Since the memory has a physical limitation so the number of pictures that can be displayed must be limited also(say dispalying a jpeg image of 400KB wich after decompression will be atleast 1MB(considering jpeg2k which has a compression ratio 1:110).If my available ramsize is 100MB then maximum number of pictures that can be displayed is 100 ,unless it uses the HDD as buffer)).Is that the case?Please Help.



@Indi
Quote:
That requires OCR, optical character recognition software. That is very complicated, using edge detection and pattern recognition algorithms. It is unrealistic to expect such a thing to be written in any web-based language.


I know that is a part of ocr and I have an ocr engine installed in server.Now from the client, I want to send only the text part so an edge detection technique is required.Which language will be suitable for that? PHP,JSP or .NET(Vb or VC)?Please help
Indi
bukaida wrote:
Quote:
It must always be decompressed. Sometimes that's done in hardware, but it must always be done.

If that be the case, then where does it store the decompressed picture?Since the memory has a physical limitation so the number of pictures that can be displayed must be limited also(say dispalying a jpeg image of 400KB wich after decompression will be atleast 1MB(considering jpeg2k which has a compression ratio 1:110).If my available ramsize is 100MB then maximum number of pictures that can be displayed is 100 ,unless it uses the HDD as buffer)).Is that the case?Please Help.

Yup, that's true. Buuuut... the JPEG compression ratio is highly variable. The best way to determine how much space an image will take up when uncompressed is by looking at the dimensions and the colour depth. An image that is 1280 × 1024 × 32 bits will take up 41,943,040 bits or 5,242,880 8-bit bytes, no matter how it was compressed.

Even low-end computers today have 256 MiB of RAM, which means that you could have roughly 50 of those images in memory at the same time on a low end computer. That's not bad at all. And generally speaking, you start with at least 2× RAM's worth of "virtual" memory - that is, hard drive space allocated for swapping - so even the lowest end computer you can find today can keep roughly 150 1280 × 1024 × 32 bit images open at the same time - without any application-assisted buffering.

So unless you're doing serious crazy image editing - in which case you should really have a higher end computer anyway - it's not really a big deal. Hell even my crap cell phone has plenty of memory, and the images it can display are tiny.

bukaida wrote:
@Indi
Quote:
That requires OCR, optical character recognition software. That is very complicated, using edge detection and pattern recognition algorithms. It is unrealistic to expect such a thing to be written in any web-based language.


I know that is a part of ocr and I have an ocr engine installed in server.Now from the client, I want to send only the text part so an edge detection technique is required.Which language will be suitable for that? PHP,JSP or .NET(Vb or VC)?Please help

If you already have an OCR engine, then you already have the edge detection routines built in to it. As for how to invoke your OCR software, you're going to have to RTFM.
bukaida
Quote:
If you already have an OCR engine, then you already have the edge detection routines built in to it. As for how to invoke your OCR software, you're going to have to RTFM.


My OCR engine is written in VC++ and hence doesnot support multiuser environment.Please help me with the idea to extend it to a client server environment.Many many thanks in advance.
bukaida
Anybody having any idea about running an ocr in client server environment?
Reply to topic    Frihost Forum Index -> Scripting -> Others

FRIHOST HOME | FAQ | TOS | ABOUT US | CONTACT US | SITE MAP
© 2005-2007 Frihost, forums powered by phpBB.