Understanding Bit Types
From Isis Workshop
Contents |
[edit] Bit Type Basics
Introduction to Bit Types
Computers store values in base-2 or binary (see glossary below). Values are generally limited to 8, 16, or 32 binary digits, or bits. We refer to the number of bits per value as the bit type. Understanding the bit type and related attributes of an Isis image cube is important for a variety of reasons. For instance, changing the bit type affects the file size, and a cube's bit type determines the range and number of values that can be stored in a cube.
There are three bit types utilized in Isis:
| Bit Type | Name | Number of Bytes | Range |
|---|---|---|---|
| 8 | Unsigned Byte | 1 | 0 to 255 |
| 16 | Signed Word | 2 | -32768 to 32767 |
| 32 | Real | 4 | -3.40282E+38 to 3.04282E+38 |
As shown in the illustration below, there are 8 bits in 1 byte, and 16 bits in 2 bytes. A 16-bit value is called a word.
There are 256 possible values (called digital numbers or DNs) that can be represented in an 8-bit file. If all the bits above are set to 1, the output DN value would be 255. The values 0 to 255 are derived by setting different bit positions to 0 or 1. The example below shows the binary number stored by the computer, and the corresponding DN value the user would see on a computer monitor.
| Binary Number | DN Value |
|---|---|
| 00000000 | 0 |
| 00000001 | 1 |
| 00000010 | 2 |
| ... | ... |
| 11111110 | 254 |
| 11111111 | 255 |
As you increase the number of bits, the number of possible DN values that can be represented also increases.
[edit] Glossary & Review
- Binary: Refers to the base-2 numeral system. For an indepth discussion of the binary numeral system, see this Wikipedia encyclopedia entry for Binary numeral system.
- Bit: Is short for binary digit, which in a computer is the smallest unit of storage. Bits are either "0" or "1".
- Byte: Is short for binary term. A byte is a collection of computer bits. On many modern computers, a byte is a group of eight bits. Typically, computers handle and store information in binary (base-2), and a group of eight bits represents an eight digit binary number.
- Bit type: Refers to how many bits there are per single meaningful value in an image cube file.
- DN: An abbreviation of digital number. For images, particularly Isis cubes, a DN is also referred to as a pixel.
[edit] What's the bit type of my image?
The keyword Type in the Pixels group of an image label will be set to UnsignedByte for 8-bit cubes, SignedWord for 16-bit cubes, or Real for 32-bit cubes. The figure below shows a portion of an 8-bit cube label highlighting the bit type.
Object = IsisCube
Object = Core
...
Group = Pixels
Type = UnsignedByte
ByteOrder = Lsb
Base = -128.00395256917
Multiplier = 1.00395256917
End_Group
End_Object
End_Object
[edit] How does the bit type affect my file?
The file size for the output file increases as the bit type increases because each pixel in the data portion of the file requires an increasing number of bytes to hold the value of that pixel, regardless of what that value is. While the data portion of the cube is the lion's share of the file size, the text portions of a cube (the labels and history) can add substantially to the file size. The minimum label size is 64 kilobytes. The following sizes are for a file that is 100 samples by 100 lines (click links under Bit Type to download the example cube):
| Bit Type | Data Size (bytes) | Label & History Size (bytes) | File size (bytes) | Possible DNs | DN Range |
|---|---|---|---|---|---|
| 8-bit | 10,000 | 72,348 | 82,348 | 256 | 0 to 255 |
| 16-bit | 20,000 | 78,736 | 98,736 | 65,535 | -32768 to 32767 |
| 32-bit | 40,000 | 91,507 | 131,507 | ≈ 4 billion | -3.40282E+38 to 3.40282E+38 |
[edit] How big will my cube file be?
The following simple calculator will show you how big the data portion of a single band cube file will be. Keep in mind, there a numerous factors(such as labels and additional bands) that can increase the file size of your cube. However, the data portion will usually represent the majority of the file size.
[edit] Image data size Calculator
[edit] How do I set the bit type?
Many of the ISIS programs allow the user to set the bit type by setting the bit type attribute for an output file.
| Number of Bytes | Output Bit Type |
|---|---|
| 1 | 8-bit |
| 2 | 16-bit |
| 4 | 32-bit |
[edit] Increasing the Bit Type of a Cube
Increasing the bit type of a cube is a fairly straightforward matter because the range of data in a smaller bit type cube fits easily in the increased range offered by the larger bit type.
Example - create a 16-bit (signed word) cube from an 8-bit (unsigned byte) cube using cubeatt by appending the attributes you wish to change to the file name:
cubeatt from=8bit.cub to=16bit.cub+SignedWord
OR
cubeatt from=8bit.cub to=16bit.cub+16-bit
In the cubeatt graphical user interface (GUI), simply hit the ATT (attributes) button next to the output file selection box and select the desired bit type in the Attributes dialog. Notice when you change the attributes through the ATT dialog, the attributes are appended to the output filename in the main application window in the same fashion as the command line. Most Isis applications allow you to change key cube attributes through the Attributes dialog.
[edit] Decreasing the Bit Type of a Cube
Decreasing the bit type of a cube is a bit trickier because the range of data in a larger bit type cube probably does not fit easily in the decreased range of the smaller bit type. In this case, you must supply the minimum and maximum values of the input file to convert to valid DNs in the output file. The values will then be stretched (scaled) as necessary to fit into the reduced data range. To perform this operation, you must know the range of the input data and provide that information to the application that is used to reduce the bit type.
1. Get statistics for your image: Run stats to get the data range:
stats from=32bit.cub
stats produces:
Group = Results From = 32bit.cub Average = 3386367.6610505 StandardDeviation = 7446.038790893 Variance = 55443493.675484 Median = 3385719.8387097 Mode = 3382280.056696 Skew = 0.26100683557387 Minimum = 3372788.0 Maximum = 3417331.0 TotalPixels = 265420800 ValidPixels = 265420800 NullPixels = 0 LisPixels = 0 LrsPixels = 0 HisPixels = 0 HrsPixels = 0 End_Group
2. Change the bit-type attribute of your image: Run cubeatt, using the minimum and maximum values from stats (refer to the output from stats above) for the data range attribute. Append the attribute 3372788.0:3417331.0 to the output filename create the 16-bit cube:
cubeatt from=32bit.cub to=16bit.cub+UnsignedWord+3372788.0:3417331.0 or cubeatt from=32bit.cub to=16bit.cub+16-bit+3372788.0:3417331.0
3. Voila! The resulting output cube will have the values from 3372788.0 to 3417331.0 from the original cube scaled to the range of -32752 to 32767. Note the use of -32752 instead of -32768. This is because the values below -32752 to -32768 are reserved for special pixels in Isis
[edit] Tips
- Increasing the bit type of a cube will increase the file size.
- When reducing the bit type, the original values may be stretched (or scaled) to fit in the range of the target bit type. This may not only shift the DN values to the new range, but may actually merge ranges of DNs into a single value if the number of distinct values in the original file is greater than the range of the output bit type.
