Photogrammetry Final Notes

3 ways to get tree heights

photogrammetry

Lidar

InSAR

2 types of sensors in cameras

CCD

CMOS

Bayer Pattern

Checkerboard filter in any camera

Each pixel has all 3 colors but filter allows it to just see 1

brightness value vectors of length 1

50% green, 25% red, 25% Blue

Subtractive Primaries

Yellow, Magenta, Cyan

Mixing RGB with equal intensity

Pigments

minus blue filter

appears yellow ( mix of red and green)

constrains blue light to see R,G, and NIR

false color composite

eliminates Rayleigh scattering

No diffuse illumination (deeper, darker shadows)

what paint color degrades the fastest?

red, it absorbs green and blue light which have more energy

aperture

how big the hole is

smaller aperture → increased depth of field

(squinting to see farther)

shutter speed

how long do you let light come in

shorter can allow us to stop action/eliminate blur

Additive Primaries

RGB

light

Yellow

Red + Green

Larger F-stop

smaller aperture

focal length (f)

distance from the lens at which parallel light rays come to a point

field of view

variable → zoom camera

fixed → can’t zoom change

F-stop (F)

Focal length/diameter

controls depth of field through changing aperture

object distance (o)

distance of camera to what your taking a picture of

image distance (i)

distance between the sensor and the lens

focus

Len’s Makers Equation

(1/F) = (1/o) + (1/i)

far away object → (1/o) goes to zero

→image distance = focal length (no need to focus)

ISO

sensitivity to light

larger → greater sensitivity → more noise

lower → more light needed

SNR

signal to noise ratio

increase signal by getting more photons in

(how wide and long lens is open)

faster shutter speed

to capture something moving fast

to capture form a fast moving plane

Things you can change in a camera

ISO - detector sensitivity

F-stop - depth of field

i - focus

f- field of view (zoom)

t - stop action

Exposure equation

E=(seen brightness*shutter speed)/ (4*F-stop²)

Increased focal length

exposure decreases

wildlife camera

you lose a lot of exposure because f is squared

why you would need a tripod

parallax

apparent displacement in the location of an object caused by a change in view position

finger moving in front of face

closer object → more

need to know flight line

changes with elevation

parallax in human eye

we stop getting it at around 50 feet

control points

provides reference system to known location

endlap

overlap between adjacent photos along flight line

stereo coverage (need at least 50% overlap)

sidelap

overlap between photos in adjacent flight lines

coverage

Lidar

Light Detection and Ranging

active

calculates distances using measured times

uses NIR or green for laser pulses

gives x,y,z locations

can travel through canopy

lidar footprint

area on ground covered by an individual pulse

lower flying height = smaller footprint= finer resolutions

waveform lidar

get intensity of entire waveform

can detect ground with high accuracy

amount of reflected energy

more data

discrete lidar

point clouds

values at peak of waveform

2-5 returns

saves time and space

scanning

push broom

profiling

single beam and going forward

most satellites use this

pulse density/frequency

number of pulses per unit on the ground

fly lower and slower, or more passes for more

lower → less likely to hit top of trees

can lower and still get similar idea for canopy distrubution

lidar intensity

how much NIR is reflected back to the sensor

using amount of photons

can use to illustrate clouds

mosaic

stitched together flightlines

DSM (digital surface model)

put a blanket over an area

elevation of top height

like a DEM but includes both manmade and natural objects

air base (B)

distance plane flew between photos

image cross correlation

easier in well defined areas (road or stream intersections)

can lead to better height quality

H’

flight height above ground

H

flying height above mean sea level

h

elevation of ground surface

scale (S)

changes with elevation

(distance on map/distance on ground)

scale equations

S=d/D=f/H-h=P/B

structure from motion

creating a 3D model from connecting multiple pictures

requires more overlap than photogrammetry

beam divergance

uses mrad to calculate

0.7 mrad at 1000m = 0.7 m footprint diameter

3 ways to collect lidar data

ground

airplane

space

lidar return

energy returned back to sensor

distance formula

d= (travel time x speed of light)/2

ground elevation

altitude - distance

orientation of plane

inertial measurement unit

canopy height model

DSM-DEM

sensor dead time

your sensor has to reset after a pulse

cannot make reading during this time

may prevent you from seeing the ground

leads to underestimations of tree height

segmentation

can combine data types to get structure and function

can determine chlorophyll count, ect…

fires

have peak spectral radiance at around 4 micrometers

lots in Canada, Alaska, and Siberia

fire band

narrow

placed onto sensor of satellite

at around 4 micrometers

can see fires within minutes

differential normalized burn ratio

used to detect fire severity

largest difference between veg and burned area at NIR and SWIR2

SWIR2 is harder to detect due to less energy in EM spectrum

high means high regrowth

low means severe fire

fuel bed

homogeneous unit that will have similar burn characteristics

remote sensing options

government moderate to low resolution satellites

commercial high spatial resolution

crewed aircraft

uncrewed aircraft (drone)

Simple ratio

(NIR/R)

Way to distinguish leaves in an area

larger indicates more healthy vegetation