lecture 7 digital image

Question 1

What does a DN represent

Answer 1

represents the radiance measured in that pixel in that spectral band. The range of DN values used depends on radiometric resolution of the sensor. Reflectance from the surface. This number gets converted into a quantitative unit.

Question 2

how do you interpret an image

Answer 2

Tone – brightness of the image, soil – farming/land use if the fields are ploughed or not.
Shape – outline of object,
Pattern – tree crowns for example – farming/plantation
size – size of the road would indicate the type of the road – also indicate function
Texture – distinguish vegetation – forest canopy from other plants
Shadow – can see heights, terrain of mountainous areas.

Question 3

how are images displayed (DN)

Answer 3

one-to-one relationship between input brightness values and the resultant intensities of the output brightness values on the display. * For example, for a 8 bit image an input DN of 0 would result in a very dark (black) intensity on the output display, while a DN of 255 would produce a bright (white) intensity. * For an 10 bit image an input DN of 0 would result in a very dark on the output brightness map, while a DN of 1023 would produce a white output.
areas under cloud, very dark, brightest are clouds. variation in brightness in an image can influence how it is displayed.

Question 4

How can RGB be displayed?

Answer 4

▪ True colour composite
▪ False colour composite (FCC)
The aim of displaying images using particular waveband combinations is to enhance the display of certain features in the image – so you can interpret it – highlight the features you are interested in.

Question 5

What does an image histogram do?

Answer 5

An image histogram is a graphical representation of the number of pixels (frequency) in an image as a function of their Brightness values (DN)

Question 6

If the histogram had lots of lines to the left of the graph…

Answer 6

Dark image, values that are very low, lots of pixels have no reflectance

Question 7

If the histogram has image values very packed together (example shows them in the middle)

Answer 7

low contrast

Question 8

If histogram shows the pixels across the right of the histogram

Answer 8

Normal interpretation of the image

Question 9

what does image enhancement do

Answer 9

Image enhancement aims to improve how the image is DISPLAYED on our screen – aimed to help us analyse the data ▪ by using the entire brightness range of the display ▪ not changing the values in the image

Question 10

What is contrast stretching

Answer 10

▪ Expands the original input brightness values to make it the same as the total dynamic range or sensitivity of the screen. Improve scaling of data

Question 11

what are the types of contrast enhancement

Answer 11

▪ Linear
▪ Histogram equalisation
▪ Gaussian stretch

Question 12

Linear contrast stretch

Answer 12

▪ Pixel values are linearly scaled
▪ Lowest image value -> Lowest displayed value (0)
▪ Highest image value -> Highest displayed value (255)
All of the data below 84 and above 153 is not being used, each colour displaying between 0 and 255.
improves the contrast of the image, distributes the values over a wider range

Question 13

Why do we need data pre processing

Answer 13

▪ To remove deficiencies in the data comparisons would be in error if not corrected, such as one image being covered with Sarah dust. satellites are orbiting the earth.

Question 14

▪ When do we need to pre-process data?

Answer 14

▪ Comparing images acquired at different dates or times - e.g., physical properties, cover types
▪ Deriving properties that can only be estimated by reflectance characteristics (i.e., quantitative analysis)

Question 15

▪ What deficiencies do we need to remove from the data?

Answer 15

▪ Radiometric fidelity – data itself
▪ Atmospheric perturbations – atmosphere interference
▪ Geometric error – coordinates for the earth surface may have to be mapped – geometric correction and can incur errors

Question 16

how are the errors removed

Answer 16

1. Radiometric calibration - converts digital number from value to temperature unit
2. Atmospheric correction - influence of atmosphere removed from data.
3. Geometric correction - satellite image - relate different data sets to the image to use GIS to carry out additional analysis. Landsat data remapped to surface. some satellite data you perform this yourself.

Question 17

What is radiometric calibration

Answer 17

▪ Satellites measure spectral radiance (W∙m-2 )
▪ describes the amount of energy reflected or emitted from a surface area
▪ this is converted to digital number onboard the satellite
▪ Radiometric calibration is the process of converting DN to radiance, reflectance or brightness temperature

Question 18

Why calibrate?

Answer 18

To make use of the imagery in a quantitative manner
▪ estimate physical properties of the surface
▪ E.g., vegetation health, surface temperature, water quality
▪ To compare information from different sensors (e.g. Landsat TM and SPOT HRV) or different time periods
▪ Needed for consistent time-series of data
▪ Advanced Very-HighResolution Radiometer (AVHRR) has been operating since 1980 so not so advanced now
▪ Important for climate change studies
▪ Satellites drift in their orbit
▪ Alters image acquisition time – therefore the measurements
want consistent data – consistent sun in the sky.

Question 19

What is a true colour image?

Answer 19

red is displayed in the red waveband, green in green and blue in blue

Question 20

What is an image histogram?

Answer 20

Graph that displays the frequency distribution of image pixel values. Would do this to enhance the image, contrast is improved as histogram stretched across full range.