Image Interpretation and Analysis

講義予定 Lecture Schedule
都市基盤工学（リモートセンシングとGIS）
(1) 2016年 4月13日（水）イントロダクション Introduction
(2) 2016年 4月20日（水）リモセンの基礎原理1 Fundamentals of RS #1 Liu
(3) 2016年 4月27日（水）リモセンの基礎原理2 Fundamentals of RS #2
(4) 2016年 5月 2日（月）衛星とセンサ1 Satellites and sensors #1
(5) 2016年 5月11日（水）画像解析1 Image Analysis #1 Liu
(6) 2016年 5月18日（水）画像解析2 Image Analysis #2 Liu
(7) 2016年 5月25日（水）衛星とセンサ2 Satellites and sensors #2
(8) 2016年 6月 1日（水）衛星とセンサ3 Satellites and sensors #3
(9) 2016年 6月 8日（水）マイクロ波リモセン#1 Microwave RS #1
(10) 2016年 6月 15日（水）GISの基礎1 Basics of GIS #1 Maruyama
(11) 2016年 6月22日（水）GISの基礎2 Basics of GIS #2 Maruyama
(12) 2016年 6月29日（水）GISの基礎3 Basics of GIS #3 Maruyama
(13) 2016年 7月 6日（水）課題発表1 Presentation by Students #1
(14) 2016年 7月 13日（水）課題発表2 Presentation by Students #2
(15) 2016年 7月27日（水）課題発表3 Presentation by Students #3
Remote Sensing and GIS
第5回
2016. 5. 11
千葉大学大学院工学研究科
Graduate School of Engineering, Chiba University
建築・都市科学専攻都市環境システムコース
Department of Urban Environment Systems
劉ウェン Wen LIU
http://ares.tu.chiba-u.jp/
2
Image Interpretation (1)
画像判読
Image Interpretation and Analysis
画像の判読と処理
 Interpretation and analysis of remote sensing
imagery involves the identification and/or
measurement of various targets in an image in
order to extract useful information about them.
リモセン画像から解析対象物を決める
Visual interpretation 目視判読
Digital processing デジタル画像処理
Preprocessing 前処理
Image Enhancement 画像強調
Image Transformation 画像変換
Image Classification & Analysis 画像分類・解析
Integration 統合
3
Visual interpretation and analysis dates back to the
early beginnings of remote sensing for air photo
interpretation. Digital processing and analysis is more
recent with the advent of digital recording of remote
sensing data and the development of computers.
目視判読からデジタル画像処理へ
Recognizing targets is the key to interpretation and information
extraction. Observing the differences between targets and their
backgrounds involves comparing different targets based on any, or all,
of the visual elements of tone, shape, size, pattern, texture, shadow,
and association.
色調，形，大きさ，パターン，テクスチャ，影，付随物
4
Image Interpretation (2)
Image Interpretation (3)
Tone refers to the relative brightness or color of
objects in an image. Generally, tone is the fundamental
element for distinguishing between different targets or
features. Variations in tone also allows the elements of
shape, texture, and pattern of objects to be
distinguished. 色調は相対的な明るさや色
Shape refers to the general form, structure, or
outline of individual objects. Shape can be a very
distinctive clue for interpretation. Straight edge shapes
typically represent urban or agricultural (field) targets,
while natural features, such as forest edges, are
generally more irregular in shape.
形は識別に重要，人工物は整形，自然は不整形
Size of objects in an image is a function of scale. It is
important to assess the size of a target relative to other
objects in a scene, as well as the absolute size, to aid in
the interpretation of that target.
大きさは対象の識別に重要
5
Digital Image Processing
Pattern refers to the spatial arrangement of visibly
discernible objects. Typically an orderly repetition of similar
tones and textures will produce a distinctive and ultimately
recognizable pattern. パターンは識別できるものの配置
Texture refers to the arrangement and frequency of tonal
variation in particular areas of an image. Rough textures
would consist of an irregular tone where the grey levels
change abruptly in a small area, whereas smooth textures
would have very little tonal variation. テクスチャは色調の変化
Shadow is also helpful in interpretation as it may provide
an idea of the profile and relative height of a target or targets
which may make identification easier.
影は高さ等の識別を容易にする
Association takes into account the relationship between
other recognizable objects or features in proximity to the
target of interest. The identification of features that one
would expect to associate with other features may provide
information to facilitate identification.
付随物は対象の識別を容易にする
6
Preprocessing 前処理
デジタル画像処理
 In order to process remote sensing imagery
digitally, the data must be recorded and available
in a digital form suitable for storage on a
computer tape or disk.
デジタル画像処理にはデジタルデータ
 Preprocessing functions involve those operations that are normally
required prior to the main data analysis and extraction of information,
and are generally grouped as radiometric or geometric corrections.
前処理には放射量補正と幾何補正がある
Today virtually all image interpretation and analysis involves digital
processing. Several commercially available software systems have been
developed specifically for remote sensing image processing and analysis.
市販の画像解析用ソフト
Most of the common image processing functions available in image
analysis systems can be categorized into the following four categories:
Preprocessing 前処理
Image Enhancement 画像強調
Image Transformation 画像変換
Image Classification & Analysis 画像分類・解析
7
Radiometric corrections include correcting the data for sensor
irregularities and unwanted sensor or atmospheric noise, and converting
the data so they accurately represent the reflected or emitted radiation
measured by the sensor.
放射量補正は，センサや大気の不整・ノイズを除く
Geometric corrections include correcting for geometric distortions
due to sensor-Earth geometry variations, and conversion of the data to
real world coordinates (e.g. latitude and longitude) on the Earth‘s
surface.
幾何補正はセンサと地球の位置関係などの歪みを是正
8
Radiometric corrections (2) Atmospheric Correction
Radiometric corrections (1) 放射量補正
周辺減光
Randomness of sensitivity
センサ補正
A sensor will receive not only the direct reflected or emitted radiation
from a target, but also the scattered radiation from a target and the
scattered radiation from the atmosphere, which is called path radiance.
Atmospheric correction is used to remove these effects.
The atmospheric correction
method is classified into the
method using the radiative
transfer equation, the
method using ground truth
data and other methods.
太陽高度・地形補正
大気補正
大気補正は放射伝達方程式を利用す
るか，地上観測値を利用する
直達光・反射光の減衰
天空光：大気によって
散乱された光
光路放射：
センサに届く、対象物からの反射，放射以外に大
気によって散乱された光
Vignetting: With the use of a lens, a fringe area in the corners
becomes darker as compared with the central area.
周辺減光：光学系レンズで中心部より周辺が暗くなる現象
9
Geometric Distortions 幾何学的ひずみ
走査線のひずみ
縮尺誤差
射影ひずみ
縦横比のひずみ
地球曲率によるひずみ
中心投影方式のセンサの外部ひずみ
多重反射
Geometric Registration
Geometric distortion is an error on an image, between the actual image
coordinates and the ideal image coordinates which would be projected
theoretically with an ideal sensor and under ideal conditions.
平行移動
反射光の散乱
光学センサの入射光の
大気による影響
10
位置合わせ
 To correct unsystematic or random errors, geometric registration of
the imagery to a known ground coordinate system must be performed.
幾何的なランダム誤差を除くには地球座標系との位置合わせが必要
The geometric registration process involves identifying the image
coordinates of several clearly discernible points, called ground control
points (or GCPs), in the distorted image (A - A1 to A4), and matching
them to their true positions in ground coordinates．地上基準点
The true ground coordinates are typically measured from a map (B B1 to B4), either in paper or digital format. This is image-to-map
registration. 画像の地図への位置合わせ
 Geometric registration may also be
performed by registering one images to another
image, called image-to-image registration,
and is often done prior to performing various
image transformation procedures.
多時期画像を用いる時などは画像同士の位
置合わせが必要
斜めひずみ
地形起伏による
ひずみ
11
12
Image-to-Image registration for change detection
Resampling and Interpolation (1) 再配列と内挿
２時期画像の変化抽出
 In order to geometrically correct the original distorted image, a
procedure called resampling is used to determine the digital values to
place in the new pixel locations of the corrected output image. The
resampling process calculates the new pixel values from the original
digital pixel values in the uncorrected image.
元の幾何的に歪んだ画像を修正するには，
再配列が必要．
Pre-event
Post-event
Master image
Master image
Slave image
GCP: Well defined shapes, like road
intersections, border of dams,
distinctive water bodies, rivers, etc.
(i,j)
(i,j+1)
u=f(x,y)
v=g(x,y)
(i+1,j)
Slave image
i’
(i+1,j+1)
j’
13
There are three common methods for
resampling: nearest neighbor, bilinear
interpolation, and cubic convolution.
3つの再配列法
Nearest neighbor resampling uses the
digital value from the pixel in the original image which is nearest to
the new pixel location in the corrected image. This is the simplest
method and does not alter the original values, but may result in some
pixel values being duplicated while others are lost. This method also
tends to result in a disjointed or blocky image appearance.
最近隣内挿法：最も近いピクセルの値に置き換える方法
14
Nearest-neighbor
Resampling and Interpolation (2) 再配列と内挿
Bilinear interpolation resampling takes a
weighted average of four pixels in the original
image nearest to the new pixel location. The
averaging process alters the original pixel
values and creates entirely new digital values in
the output image. This may be undesirable if
further processing and analysis is to be done.
共１次内挿法は４つのピクセルの値の線形式
Cubic convolution resampling goes even further
to calculate a distance weighted average of a block
of sixteen pixels from the original image which
surround the new output pixel location. As with
bilinear interpolation, this method results in
completely new pixel values. However, these two
methods both produce images which have a much
sharper appearance and avoid the blocky
appearance of the nearest neighbor method.
3次たたみ込み内挿法は16ピクセルの３次式に
よる内挿．鮮鋭化の効果あり
Interpolation methods 内挿法
15
Remote Sensing Note, Japan Association on Remote Sensing, 1996.
16
Area correlation of Two Images in Sub-pixel Level
Synthetic Middle-Resolution Images and Offset
52 pixels x 28.8 m  1.5km
Low resolution image
Reference image
High resolution image
Template
From: 2.4m
To: 28.8 m
Offset
N
K
M
Assumed Point Spread
Function (PSF)
Synthetic Middle-Resolution Image
with different offset
Correction Matrix
1.0
Coefficient of determination
L
i
Coefficient of Determination vs. Offset
0.9
Band1
0.8
Band2
Band3
Band4
0.7
0.6
j
0.5
M
rij 
0.4
0.3
0.2
0.1
0.0
2.4
0
7.2
5
14.4
10
21.6
15
20
28.8
25
30
Offset (m)
N
 (T
m 1 n 1
m,n
 T )( Si  m, j  n   S )
1/ 2
M N
 M N

 (Tm,n  T )  ( Si  m, j  n   S )
 m 1 n 1
  m 1 n 1

17
Registration – Location of the GCP’s
(Maximum Value of the Spatial Cross-correlation)
Sub-scene from
the target image
Sub-sampled up to
1/12
(cubic convolution)
624 pixels
200 pixels
Cross-Correlation of 7.2m Offset - Template 08
0.8
100
0.6
200
300
0.4
400
0.2
500
0
600
-0.2
700
624 pixels
-0.4
800
Correlation matrix showing the maximum cross-correlation
100
200
300
400
500
600
700
800
By Dr. Miguel Estrada
18
Image Enhancement (1) 画像強調
Location of the offset by spatial cross-correlation index
200 pixels
Reference image
1/ 2
pixel: 302
line: 301
19
 Image enhancement is solely to improve the appearance of the
imagery to assist in visual interpretation and analysis. Examples of
enhancement functions include contrast stretching to increase the tonal
distinction between various features in a scene.
判読や解析を容易にするための画像色調の改善
The advantage of digital imagery is that it allows us
to manipulate the digital pixel values in an image.
Although radiometric corrections for illumination,
atmospheric influences, and sensor characteristics may
be done prior to distribution of data to the user, the
image may still not be optimized for visual
interpretation. デジタル画像は数値の操作が容易
With large variations in spectral response from a diverse range of
targets (e.g. forest, deserts, snowfields, water, etc.) no generic
radiometric correction could optimally account for and display the
optimum brightness range and contrast for all targets. Thus, a custom
adjustment of the range and distribution of brightness values is usually
necessary.様々な対象物に対応するには，適宜，明度の修正が必要 20
Image Enhancement (2)
 In raw imagery, the useful data often populates
only a small portion of the available range of
digital values. 可能な輝度値の範囲の一部しか
使っていない場合が多い.
Image Enhancement (3)
8-bit image (0 - 255
brightness levels)
Contrast enhancement involves changing the
original values so that more of the available range is
used, thereby increasing the contrast between
targets and their backgrounds.
コントラスト強調はより広い輝度値の範囲に広げる
A histogram is a graphical representation of the brightness values
that comprise an image. The brightness values are displayed along the
x-axis of the graph. The frequency of occurrence of each of these
values in the image is shown on the y-axis.
ヒストグラムは画像の輝度値の頻度分布を示す
By manipulating the range of digital values, we can apply various
enhancements to the data. The simplest type of enhancement is a linear
contrast stretch. This involves identifying lower and upper bounds from
the histogram and applying a transformation to stretch this range to fill
the full range. 最も単純なコントラスト強調は線形ヒストグラム伸張
輝度値のヒストグラム
x-axis = 0 to 255
y-axis = pixel数 21
After Stretch
22
Image Enhancement (4)
Example of Linear Contrast Stretch
A uniform distribution of the input range of values across the full range
may not always be an appropriate enhancement, particularly if the input
range is not uniformly distributed. In this case, a histogram-equalized
stretch may be better.
輝度値が均一分布でないときは頻度を平滑
化するような伸張も考えられる
画像表示例：明るさの自動調整（多バンド）
Funabashi, Chiba Prefecture, Landsat TM 1984 Band 1,2,3 True color
 The histogram-equalized stretch assigns more display values (range)
to the frequently occurring portions of the histogram. In this way, the
detail in these areas will be better enhanced relative to those areas of the
original histogram where values occur less frequently.
ヒストグラム平滑化により頻度の高い
領域をより詳細に見ることが可能
Linear stretch
線形ストレッチ
「はじめてのリモートセンシング－地球観測衛星ASTERで見る」，山口靖・八木令子・小田島高之監修（ジオ
テクノス発行），4000円，CD-ROM付き，2004年．
Before Stretch
23
24
ヒストグラム変換
Histogram Conversion
Histogram equalization
ヒストグラム平滑化
Example of Histogram Equalization
画像表示例：ヒストグラム平滑化による明るさの調整
Histogram normalization
ヒストグラム正規化
Funabashi, Chiba Prefecture, Landsat TM 1984 Band 1,2,3 True color
ヒストグラム
ヒストグラム
Histogram equalization
ヒストグラム平滑化
原画像のヒストグラムを正規分布に
なるように変換
累積ヒストグラム
原画像のヒストグラムを頻度が一定になる
ように変換
25
Correction of Shadow from Digital Images
11:30
13:30
15:30
日向白板反射量　(12/04)
Under Sunshine
日向
12:30
14:30
16:00
In the 日影　白板反射量
shadow (12/04)
日影
160
B
140
G
R
画像における影補正
11:30
13:30
15:30
12:30
14:30
16:00
26
Shadow Correction of VHR Satellite Images
Original image
高解像度衛星画像の影補正
Corrected shadow-free image
35
NIR
B
30
G
R
NIR
120
反
射
量
100
Irradiance
Irradiance
反
射
量
80
60
25
20
15
10
40
20
5
0
350
450
550
650
750
850
950
0
1050
350
Wavelength [nm]
Wavelength (nm)
450
550
650
750
850
950
1050
Wavelength [nm]
Wavelength
(nm)
Ratio= Shadow/Sunlit
日影/日向 (12/04)
100%
B
80%
G
R
11:30
13:30
15:30
12:30
14:30
16:00
950
1050
NIR
60%
40%
20%
0%
350
450
550
650
750
Wavelength (nm)
Wavelength [nm]
850
•Various darkness in shadow.
•The intensity ratio of shadow
and sunlit areas are time-,
season-, and location-dependent.
•The ratio is also wavelengthdependent.  Even the NDVI is
time-dependent.
27
■■■
Object-Based Shadow Extraction and Correction of High-Resolution Optical Satellite Images,
W. Liu, F. Yamazaki, IEEE JSTARS, 8, pp. 1-7, 2012.
Shadow and non-shadow pairs
28
Free Software for Image Analysis
Free Satellite Images (Landsat)
画像処理の無料ソフト
• 「RSP Ver2.03」
http://www.ctie.co.jp/software/rsp/#ttl06
• Download the zip file and decompress it
ダウンロードして解凍する
29
Landsat Image Searching
Band name
Landsat 4/5 (TM)
Landsat 8 (OLI)
Costal Aerosol
－
1
Blue
1
2
Green
2
3
Red
3
4
Near Infrared (NIR)
4
5
SWIR 1
5
6
SWIR 2
7
7
Panchromatic
8 (15 m)
8 (15 m))
Cirrus
－
9
Thermal Infrared (TIRS) 1
6 (120 m)
10 (60 m)
Thermal Infrared (TIRS) 2
6 (120 m)
11 (60 m)
Resolution 30 m
30
Limit the Target Area
 Obtain the free Landsat images from the USGS website
http://landsat.usgs.gov/index.php
31
• Find the Path and Row numbers for the target area
using LandsatLook Viewer
32
Searching the satellite images
Limit Date Range and Data Type
• Search the Landsat images using EarthExploxer
• Input the Path and Row number to limit the searching area
• Input the date range and data type for searching
33
Add Cloud Cover Condition
34
Data for Image Analysis
• Less than 10%
• Search the Landsat 8 image taken on March 28, 2014,
coving Kanto area.
35
36

Download Report