Special Pixels
From Isis Workshop
Contents |
[edit] Introduction to Special Pixels
[edit] What are special pixels?
Special pixels are defined to distinguish valid pixels from non-valid pixels in Isis. This is very important to scientists who evaluate the density values, and draw conclusions about the physical properties of a scene.
[edit] How are special pixels set?
There are two ways that Digital Number (DN) values are set to special pixels in Isis:
- Instrument Special Pixels These types of special pixels are set during the ingestion of the data into Isis because the acquired measurement either exceeded the dynamic range of the sensor based on its inherent properties (such as sensitivity) and settings (such as gain) or the measurement wasn't collected due to an instrument or transmission error.
- Low Instrument Saturation (LIS): The sensor registered a value too low to be measured accurately (i.e. undersaturated).
- High Instrument Saturation (HIS): The sensor registered a value too high to be measured accurately (i.e. oversaturated).
- No Data Collected (NULL): The instrument did not collect a measurement due to an sensor malfunction (such as a damaged CCD element)or an error in transmission prevented the value from being recorded.
- Representation Special Pixels: These types of special pixels are set due to being processed by Isis applications that cause the DNs to fall outside the data range of the file, either due to settings defined by the user or the effects of the algorithms within the application.
- Low Representation Saturation (LRS): The resulting DN calculated by the application fall below the possible range of values (i.e. undersaturated).
- High Representation Saturation (HRS): The resulting DN calculated by the application are higher than the possible range of values (i.e. oversaturated).
- Data Removed (NULL): NULL values are set by an application when it removes data, such as during masking or geometric warping.
[edit] Special Pixels in Isis
[edit] How many special pixel values exist in Isis?
|
There are five special pixel values defined in the Isis software system for 16-bit and 32-bit data type. Each of the five special pixels is assigned a specific value that represents a special type of invalid data. The NULL special pixel identifies pixels where either the instrument failed to collect data, or data was removed or empty pixels created during processing. The LRS (Low Representation Saturation) and HRS (High Representation Saturation) represent saturated pixels that fall outside the minimum and maximum valid range of the file respectively. The LIS (Low Instrument Saturation) values represent pixels that fall below the lowest value the instrument can accurately record. The HIS (High Instrument Saturation) values represent pixels that exceed the highest value the instrument can accurately record. |
Special Pixels in Isis Visual refers to the apparent color of a special pixel in an Isis display application, such as qview.
|
[edit] Are the special pixel values propagated between different bit types?
The special pixel values are propagated as is between 16-bit and 32-bit data types (see the chart above to see what the values are set to after conversion from one bit type to the other). If the image is converted from 16-bit or 32-bit to an 8-bit data type, then the LIS, LRS, and NULL special pixel values 0, and HIS and HRS special pixel values are set to 255.
[edit] Is the NULL special pixel defined as 0 for all bit types?
No. The DN value 0 is defined as NULL only in an 8-bit file. For the 16-bit and 32-bit data, 0 is a valid DN!
[edit] Interactive Special Pixel Demonstration
[edit] Working with Special Pixels
Q: How do I find out if there are any special pixels in my file?
A: Statistics: hist & stats
There are two programs stats and hist that will provide statistics about the pixels of an image. The output from stats looks like:
Group = Results From = specialpixel.cub Average = 4.57024100753882e-04 StandardDeviation = 3.16311013180614e-05 Variance = 1.00052657059347e-09 Median = 4.57129145213171e-04 Mode = 4.57129145213171e-04 Skew = -0.0099627697024315 Minimum = 2.85016489215195e-04 Maximum = 6.54914067126811e-04 TotalPixels = 1725440 ValidPixels = 1723452 NullPixels = 133 LisPixels = 3 LrsPixels = 1781 HisPixels = 19 HrsPixels = 52 End_Group
View your image: qview
You can also display the image on a monitor using the program "qview", and position the cursor over a very dark pixel. The DN displayed on the screen will either be "NULL", "LIS", "LRS", or a very low DN value. The "HIS" and "HRS" pixels will be displayed as white pixels on a monitor.
