2/16/2025

Fundam ent al A nalog/ V i sual Recogni t i on Elem ent s

1. Tone or Color
• Tone ref ers t o t he r elat ive br ight ness or color of object s on an image. Relat ive phot o t ones
(grey or color variat ion) could be used t o dist inguish bet ween dif f erent object s.
• For examples: Trees, Soil t ype and moist ure cont ent show t onal variat ion which are
ident if iable as dist inct object s or f eat ures.

1
 2/16/2025

2. S i z e
• Size of objects on images must be – Example
considered in the context of image scale. • apartments vs. houses
• single lane road vs. multilane
• Relative sizes among objects on images of
same scale must also be considered.

3. Shape 3. S hape
• Shape refers to the general form, configuration, or outline of individual objects. In case of
stereoscopic images, The objects height also defines its shape.
• Sometimes, shape could be so distinct that it’s the sole criteria for identification of an object
in an image.

• cultural features - geometric, distinct boundaries

• natural features - irregular shapes and boundaries
• Shape helps us distinguish old vs. new subdivisions, some tree species, athletic fields, etc.

2
 2/16/2025

3. S hape

4 . S hadows
Shadows are important to
interpreters in tow opposing
respects.
1. The outline of a shadow affords
as impression of the profile view
of objects (which aids
interpretation), and
2. Objects within shadows reflect
little light and are difficult to
discern on an image (which
hinders interpretation)

• shadows cast by some features

can aid in their i.d. Powerline
• some tree types, storage tanks, transmission
bridges can be identified in this
way towers
• shadows can also accentuate
terrain
10

3
 2/16/2025

4 . Tex t ur e
• Texture is frequency of tonal change in an image.
Texture is produced by an aggregation of unit features
that may be too small to be discussed individually on
the image, such as tree leaves or tree shadows.

• It is a product of their individual shape, size, pattern,

shadow and tone.

• It determines the overall visual “smoothness” or

“coarseness” of image features.

• As the scale of image changes, the texture of an image

progressively changes.

• Texture
• coarseness/smoothness caused by variability or
uniformity of image tone or color
• smoothness - surface features of similar height: crops,
bare fields, water, etc. Helyar Woods and points south
• coarseness - irregular surface: forest, lava flows
13

6. Pattern
• Pattern relates to the spatial arrangement of Braided pattern (river)

• The repetition of certain general forms or

relationships is characteristics of many
objects, both natural and constructed, and
gives objects a pattern that aids the image
interpretation in recognizing them.

• Pattern
• overall spatial form of related
features
• repeating simple geometric
patterns tend to indicate
cultural features
• more complex, curving
patterns tend to indicate
natural features A dendritic pattern is characteristic of
flat-lying sedimentary bedrock 16
• drainage patterns can help

4
 2/16/2025

7. Height and Depth

18

8 & 9. S i t e and A s s oci at i on

• Site refers to topographic or geographic
location, and is particularly important aid
in the identification of vegetation types.

• Association refers to the occurrence of

certain features in relation to others.

• Site
• oak on hillside, Atlantic.
white cedar in stream
corridor
• Association
• cooling towers, HT lines =>
power generating plant 19

5
 2/16/2025

10 . Res olut i on
Resolution is a broad term commonly used to describe:
• number of pixels on a display device, or
• the area on the ground  pixel in an image

These broad definitions are inadequate when describing remotely

sensed data. Four distinct types of resolution must be considered:

1. Spectral resolution: is the specific wavelength intervals that a sensor can be recorded by a
sensor (detector)

2. Spatial resolution: is the area on the ground represented by individual pixel

3. Radiometric resolution: is the range of values in each band (i.e. number of bits)

4. Temporal resolution: revisit time or how often a sensor obtains imagery of a particular area

24

a. S pect r al Res olut i on

specific wavelength intervals in the
electromagnetic spectrum that a sensor can record

Wide intervals in the electromagnetic spectrum are

referred to as coarse spectral resolution, and

narrow intervals are referred to as fine spectral

resolution

25

6
 2/16/2025

Hyper-spectral
Cube

Structure of the
hyperspectral
data cube.

26

b. S pat i al
Spatial resolution Res olut i on
• measure of the smallest object that can be resolved
by the sensor,
• the area on the ground represented by each pixel
• The finer the resolution, the lower the number.

large-scale imagery and

The terms
small-scale imagery often refer to
spatial resolution.

Scale is the ratio of distance on a map as related to the true

distance on the ground

• Large-scale in remote sensing refers to imagery in which

each pixel represents a small area on the ground, such
as SPOT data, with a spatial resolution of 10 m or 20 m.

• Small scale refers to imagery in which each pixel

represents a large area on the ground, such as AVHRR
data, with a spatial resolution of 1.1 km. 27

7
 2/16/2025

• Accurate spatial registration of at least two images is essential for digital change detection.
Ideally, the remotely sensed data are acquired by a sensor system that collects data with the
same instantaneous field of view on each date. For example, Landsat TM data collected at 30 30
m spatial resolution on two dates are relatively easy to register to one another.

• It is possible to perform change detection using data collected from two different sensor systems
with different IFOVs, for example, Landsat TM data (30 30 m) for Date 1 and SPOT HRV XS data
(20 20 m) for Date 2.

• In such cases, it is usually necessary to decide on a representative minimum mapping unit (e.g.,
20 20 m) and then resample both datasets to this uniform pixel size.

• This does not present a significant problem as long as the image analyst remembers that the
information content of the resampled data can never be greater than the IFOV of the original
sensor system.

32

• Geometric rectification algorithms

are used to register the images to a
standard map projection (Universal
Transverse Mercator for most U.S.
projects).

• Rectification should result in the

two images having a root mean
square error (RMSE) of <0.5 pixel.

• Mis-registration of the two images

may result in the identification of
spurious areas of change between
the datasets.

• For example, just one pixel mis-

registration may cause a stable road
on the two dates to show up as a
new road in the change image.
33

8
 2/16/2025

Radiometric resolution
• refers to the dynamic range, or number of possible data file values in
each band.
• Bit depth of an image (numeric range)
• Each grid cell or pixel has a range of grey values (0 to L-1), where L is the
max. intensity of a pixel.
0 1 2 3 4 . . . . . . . 2 2 2
5 5 5
3 4 5

8 bit image M ax. intensity

0 1 2 3 4 . . . . . . . 61 62 63

6 bit image M ax. intensity

34

35

9
 2/16/2025

c. Radi om et r i c Res olut i on

Bit depth of an image (numeric range)

An A-to-D conversion of the remote sensor data usually results in 8-bit brightness values
ranging from 0 to 255.

Ideally, the sensor systems collect the data at the same radiometric precision on both
dates.

When the radiometric resolution of data acquired by one system (e.g., Landsat MSS 1 with
6-bit data) is compared with data acquired by a higher radiometric resolution instrument
(e.g., Landsat TM with 8-bit data), the lower-resolution data (e.g., 6 bits) should be
decompressed to 8 bits for change detection purposes.

Ideally, the brightness values associated with both dates of imagery are converted to
apparent surface reflectance, which eliminates the problem.
36

• Absolute Radiometric Correction

- Use of an model atmosphere in conjunction with in situ

atmospheric measurements (if possible) to correct for path radiance.

• Relative Radiometric Correction

- Single image normalization using histogram adjustment

- Multiple date image normalization using regression techniques

Failure to understand the impact of various environmental characteristics on the

remote sensing change detection process can lead to inaccurate results. When
performing change detection, it is desirable to hold as many environmental
variables constant as possible.
37

10

10
 2/16/2025

c. Tem por al Res olut i on

• Repetition of a measurement with
respect to time.

• repeated coverage of same area

• Time series of data

Feb
• revisit time: the satellite revisit time
is the time elapsed between
observations of the same point on
earth by a satellite.

MODIS Terra NDVI

- Daily (temporal resolution)

Composites
- 8 day July
- 16 day
- Monthly
- Yearly
M ODIS NDVI
38

41

11

11
 2/16/2025

Spat ial, Spect r al, and Radiom et r ic r esolut ion

(var iat ion wit h in sam e sensor / sat ellit e)

8 bit
8 bit

30 m
8 bit

43

Temporal
9 bit 9 bit 9 bit
8 bit 8 bit 8 bit
8 bit 8 bit 8 bit

30 m 30 m 30 m
8 bit 8 bit 8 bit

Time t1 Time t2 Time t3

9 bit 9 bit 9 bit

8 bit 8 bit 8 bit
8 bit 8 bit 8 bit

30 m 30 m
30 m
8 bit 8 bit 8 bit

Time t4 Time t5 Time t6

44

12

12
 2/16/2025

Im age
Pr ocessing

Ras t e r and
V e ct or

Raster vs Vector Vector

Basic Elements:
• Location (x,y) or (x,y,z)
• Explicit, i.e. marked to a coordinate system
• Different coordinate system (and precision) require different
values
Raster (Grid) • e.g. UTM as integer (but large)
• Lat, long as two floating point numbers +/-
• Points are used to build more complex features

Vector Raster
(Feature)
Basic Elements:
• Extent
o #Rows
o #Columns
• Origin
• Orientation
Real-world • Resolution: pixel =grid cell
56

13

13
 2/16/2025

Raster vs Vector
Vector Raster

• use mathematical relationships between points and • uses a grid of individual pixels where each
the paths connecting them to describe an image. pixel can be a different color or shade.

• composed of pixels.
• mathematically-defined geometric shapes—lines,
objects and fills

• Loss of information with scaling

• composed of paths.
• Jagged appearance (for coarse resolution)
• Scalable (rescaling) without loss of information
• used for photographs and images

• Smooth appearance • Examples: Digital Scanner or Camera

• Examples: tif, png

• layout, type, line art or illustrations.

• Examples: svg, ai

57

Raster  Vector

Line map Raster representation Vector representation

58

14

14
 2/16/2025

59

Vector Raster
Advantages of Vector Data Structures:
Advantages of Raster Data Structures:
• Good representation of structures
• Simple data structures
• Compact • Overlay and combination of maps and remote sensed images easy
• Topology can be completely described • Some spatial analysis methods simple to perform
• Accurate graphics • Simulation easy, because cells have the same size and shape

• Retrieval, updating and generalization of graphics and • Technology is cheap

attributes possible

Disadvantages of Vector Data Structures: Disadvantages of Raster Data Structures:

• Complex Data Structures • The use of large cells to reduce data volumes means that
• Combination of several vector polygon maps through overlay recognizable structures can be lost and there can be a serious
creates difficulties loss of information
• Simulation is difficult because each unit has a different • Crude raster maps are considerably less beautiful than line maps
topological form • Network linkages are difficult to establish
• Display and plotting can be expensive, particularly for high • Projection transformations are time and computing resource
quality color consuming unless special algorithms or hardware is used.
• The technology is expensive, particularly for the more
sophisticated software and hardware
• Spatial analysis and filtering within polygons are impossible 61

15

15
 2/16/2025

Vector Raster
VECTOR FORMATS
EXAMPLES(raster data)
1. Shapefile • Temperature
Contains:
o Point, or multi-point • Pressure
o Line, or polyline • Precipitation
o Polygon
• Elevation
2. Coverage • Reflectance (photography/imagery)
Contains:
o Point, or node
o Arc, or line
o Polygon
o Label

EXAMPLES
• Administrative borders
• Linear features
• Roads

• Discrete habitat boundaries

63

The different ways of graphically displaying data encapsulated by

vector entity models raster models

64

16

16