The influence of pixel type\depth on the raster,

6863
14
02-02-2015 12:07 PM
JamalNUMAN
Legendary Contributor

The influence of pixel type\depth on the raster,

 

I couldn’t figure out the practical effect of the pixel type\depth on the raster. For example:

 

  1. What is the difference between a raster of 8 and 16 pixel size with unsigned type
  2. What is the difference between signed and unsigned raster with 16 pixel type?

Clip_71.jpg

 

Thank you

 

Best

 

Jamal

----------------------------------------
Jamal Numan
Geomolg Geoportal for Spatial Information
Ramallah, West Bank, Palestine
0 Kudos
14 Replies
DarrenWiens2
MVP Honored Contributor

This appears to be a rather old, unanswered question bumped to the top, but the practical effects of changing pixel type are file size and available values.

1.) 8-bit unsigned = values 0 - 255. 16-bit unsigned = 0 - 65535.

2.) 16-bit unsigned = 0 - 65535. 16-bit signed = -32768 - 32767.

The more digits (bits) your pixels hold, the larger the file size.

DanPatterson_Retired
MVP Emeritus

A heads up...the torrent of 'new' threads are the result of redirecting remote sensing and imagery related threads to their rightful home ... Imagery and Remote Sensing

JamalNUMAN
Legendary Contributor

Thank you guys for the useful input.

I couldn’t figure out

  • The advantage of the 8 bits over the 16 bits, for example.
  • Based on what the depth values is selected?
  • Based on what the signed\unsigned property is selected?

Clip_469.jpg

----------------------------------------
Jamal Numan
Geomolg Geoportal for Spatial Information
Ramallah, West Bank, Palestine
0 Kudos
DarrenWiens2
MVP Honored Contributor

It's based on your requirements for pixel and limitation of file storage. An 8 bit raster will be a smaller file than 16 bit, but also have less available pixel values. If you don't care about either, then the choice is irrelevant.

JamalNUMAN
Legendary Contributor

Thanks Darren for the prompt help,

Does this mean that the rasters with high depth have better “appearance” as more values are offered to the cells to better represent the reality. In other words, rasters with higher depth look clearer than those of low depth value.

Is this correct?

----------------------------------------
Jamal Numan
Geomolg Geoportal for Spatial Information
Ramallah, West Bank, Palestine
0 Kudos
DarrenWiens2
MVP Honored Contributor

Not really. You can degrade the appearance by reducing the original pixel depth, but you can't improve it just by making more values available.

You may be thinking about reducing the cell size and interpolating to artificially add intermediate values to an image. Of course, those intermediate values would have to be available, which is where pixel depth comes in.

GabrielUpchurch1
Occasional Contributor III

What Darren said.  To supplement his explanation, here is a good help resource on the topic:  Bit depth capacity for raster dataset cells—Help | ArcGIS for Desktop.

Sheen_EiseleLenon1
New Contributor II

A raster has four types of resolution - spatial, spectral, radiometric, and temporal. The pixel depth is under the radiometric resolution of a raster. It simply specifies how well the differences in brightness in an image can be perceived. The higher the radiometric resolution (pixel depth), the better small differences in reflected or emitted radiation can be measured. It also means that you'll have a larger volume of measured data.

However, unlike image data, computer monitor screen pixels can only have 256 unique RGB brightness values. Clearly, this limitation prevents the most of the data from being displayed with brightness exactly equal to their real value. Here you can use image enhancement like contrast stretching to achieve a better rendering of your raster.

So when to use 16-bit:

  • if you want to perform spectral analysis on your raster
  • if you want to create a visually appealing final product
  • if you want maximum clarity on areas covered by shadows and haze

8-bit:

  • if you prefer smaller file sizes
  • if your software has file limitations (say, can't open 16-bit rasters)
  • if you will use your raster as a backdrop
DanZimble
New Contributor II

Just a small point of clarification, not all rasters have a spectral resolution, since not all rasters are technically measuring the EM spectrum with sensors tuned to a specific set of wavelengths.  Some raster's pixel data represent categorical data, such as land cover classes, soil types, forest types, elevation classes, slope, aspect, or may even be derived from scanning a paper map.  Given that I thought it was worth making this distinction.  It might better be said that, "A remotely sensed image has...." .