Imagery in ArcGIS

Ahmed Awady

Esri

Platform Enabling Imagery Across the Enterprise

See the Earth, Find the Patterns, Share with Others

Professional

Imagery &

Geospatial

Analysts

Server

System of

Record

System of

Insight

System of

Engagement

Image

Processing

&

Analytics

Find thePatterns

Imagery

Management

See the Earth

Imagery

Widgets & Apps

Share with Others

Management

Map

ProductionAnalysis

Content

Visualization

& Exploitation

5 Key Imagery Capabilities of ArcGIS

Content Ways to See the World

Understands 25 sensor models and 57 raster formats

Drones Drone2Map for ArcGIS

2D Image Maps (Ortho Mosaics):• Map Accurate Image Maps

• Seamless stitched for base mapping

• Digital Surface Models

3D Elevation Products:• Point Cloud (LAS)

• 3D Mesh Models

• 3D Site Models (PDF)

Smart Inspection Photos:• Non-Distorted but Map Accurate

• Works for vertical and oblique photos

• Complete 3D Measurements

Overlapping

Static Aerial Photos

Tasking

Collection

Processing

Dissemination

Tasking

Processing

Analysis

Dissemination

Drone Mapping Drone Mapping and Analysis

Collection

Collection

Online Drone Mapping

Tasking

Collection

Enterprise Drone

Mapping and Analysis

TaskingEnterprise

Server

Image

Server

AGOL

• Flight Management Software

- Multi-User Configuration

- Unlimited Elastic Processing

- Advanced 3D Visualization

- Update Feature Layers

- Web Maps & Dashboards

• Flight Management Software

- Multi-User Configuration

- Unlimited Elastic Processing

- Robust Image Management

- Raster Analytics & Analysis

- Advanced 3D Modeling

- Update Feature Layers

- Web Maps & Dash Boards

- Drone Tasking App

- PC Processing

- Imagery Analysis

- Advanced 3D Modeling

- Update Feature Layers

- Web Maps & Dashboards

- Drone Tasking App

- PC Processing

- Advanced 3D Visualization

- Update Feature Layers

- Web Maps & Dashboards

Dissemination

Analysis, Exploitation, & Dissemination

ArcGIS Drone CollectionsFour capability tiers support individual users to broad organizations

Content Imagery Management is the Cornerstone

Imagery Services

OGC & KML

Mosaic Datasets:

An Imagery

Information Model

Multiple Sensors

& Formats

Web Cataloging -

Search & Discovery

Tile Cache versus Imagery ServicesImagery

Services

Imagery

Source

Files

Imagery

Service

Pros:• Best for Analytics and Analysis (original pixel values)

• Immediately available with NO preprocessing of the imagery

• Reduces data redundancy as only the original sources are used

• Can dynamically be converted into derived products on-the-fly

Cons:• Service response can be slightly slower (measured in milliseconds)

• Not ideal for 3D visualization

Tile

Cached

Pros:• Best for Visualization / 3D Visualization

• Fastest way to serve imagery you don’t want to process/change

• Ideal for static base maps

Cons:• Must preprocess all the imagery

• Must build and store all derivative image products as separates caches

• Cannot be used for image processing or analytics

• Does not support multispectral imagery band changes

Both Approaches can be used within the same architecture

Data Model The ArcGIS Mosaic Dataset

Thousands of Images

Dynamic Scaling

In a Data Model

Seamless Mosaics

Imagery Access to Metadata

Metadata for each

image is readily

accessible to end

user clients

Data Model Dynamic Image Services

Raster Functions

Raster Function Chains

Raster Functions +Function Editor

Raster Analysis

Text goes here

Imagery Tab

Text goes here

• Ortho mapping

• Georeferencing

• Access to the Raster Functions Pane & Function Editor

• Image Classification

• Processing Tools

Image Classification

The things we

extract from

imagery

Set of All Features

Features

Visible

Features

Of

Interest

Features Available

to Work with

Choose your data carefullyOptimize your data if possible

• Choose your best sensor

• Optimize for seasonality and phenomenology – timing is key

• Process effectively

- Normalize for sensor anomalies

- Minimize any variation in your datasets

Autumn

Winter

SummerSpring

SegmentationPreserving the object edges

• This technique helps

- preserve edges of objects

- Provides object specific values

• A pre-processing step

Segmentation in Pro – MS Image

Segmentation – fine detail, preview

Segementation – course detail, output

Supervised Image Classification

24

Input ImageSegmenter Segmented Image

Training Samples

ClassifierClassified Image

Accuracy

assessment

Generate training

& inspect *

Mean Shift Segmentation

Maximum Likelihood

Support Vector Machine

Random Trees

Train Test/Classify.ecd

• New in Pro 2.0

Confusion matrix

Reference Image

Classified

Image

Confusion matrix

Unsupervised Image Classification

27

Input ImageSegmenter Segmented Image

Classifier Classified Image

Accuracy

assessment

ISOData

Human Labels Reclassify

Classified Image

Imagery

Visualization & Measure

Classification & Extraction

Change Detection

Raster Analytics Cloud Processing

Elastic Processing in Scalable Environments

Image Processing – Raster Analytics

What is Raster Analytics?

• ArcGIS has a new way to create and execute spatial

analysis models and image processing chains which

leverage distributed storage and analytics

- Raster Analytics works with your existing GIS data and imagery

- register your data and go, importing existing data to distributed storage is not mandatory

- Raster Analytics can optimize your data for distributed analytics

- import your data into ArcGIS distributed storage which further improves the scalability of distributed analytics

- Raster Analytics is designed to scale with your organization’s demands

- scale up to get the job done, scale down when resources are no longer needed

Solve New Problems with Raster Analytics

• run models against data that is too big for single desktop

- small and medium scale global rasters (big geography)

- large scale local or regional rasters (high resolution)

• run models against massive collections and scale it

• run models and meet time constraints

months weeks days hours minutes

Raster Analytics is Powerful

• run a model based on a single function

• run a model by combining many functions

Math

Abs

Arithmetic

Band

Arithmetic

Calculator

Divide

Exp

Exp10

Exp2

Float

Int

Ln

Log10

Log2

Minus

Mod

Negate

Plus

Power

Round Down

Round Up

Square

Square Root

Times

Bitwise And

Bitwise Left

Shift

Bitwise Not

Bitwise Or

Bitwise Right

Shift

Bitwise Xor

Boolean And

BooleanNot

Boolean Or

Boolean Xor

Equal To

Greater Than

Greater Than

Equal

Is Null

Less Than

Less Than

Equal

Not Equal

ArgStatistics

Cell Statistics

Statistics

ACos

ACosH

ASin

ASinH

ATan

ATan2

ATanH

Cos

CosH

Sin

SinH

Tan

TanH

Data Management & Conversion

Raster to Vector

Vector to Raster

Colormap

Colormap To RGB

Complex

Grayscale

Remap / Reclass

Spectral Conversion

Unit Conversion

Vector Field

LAS to Raster

LAS Dataset to Raster

Clip

Composite

Extract Bands

Mask

Mosaic Rasters

Rasterize Features

Reproject

Interpolation

Interpolate Irregular Data

Nearest Neighbor

IDW

EBK

Swath

Correction

Apparent Reflectance

Geometric Correction

Speckle Filtering (Lee,Frost,Kuan)

Analysis: Image Segmentation & Classification

Segmentation (Mean Shift)

Training (ISO, SVM, ML)

Supervised Classification

Visualization & Appearance

Contrast and Brightness

Convolution

Pansharpening

Resample

Statistics and Histogram

Stretch

Surface Generation & Analysis

Aspect

Curvature

Elevation Void Fill

Hillshade

Shaded Relief

Slope

Viewshed

Analysis: Overlay

Weighted Sum

Weighted Overlay

Analysis: Band Math & Indices

NDVI / NDVI Colorized

SAVI / MSAVI / TSAVI

GEMI

GVI (Landsat TM)

PVI

Tasseled Cap (Kauth-Thomas)

Binary Thresholding

Analysis: Distance & Density

Euclidean Distance

Cost Distance

Cost Path

Kernel Density

Analysis: Zonal

Zonal Statistics

Zonal Histogram

Zonal Geometry

Python

Custom Algorithms

Conditionals

Con

Set Null

Raster Analytics is Easy

• easy to get started, it is “out of the box analytics”

- install on nodes -> start Raster Analytic services -> go

• ArcGIS Pro user experience

- just works with layers

- visual modeler to design simple and complex models

• results are immediately available in your Web GIS

- no publishing workflow required

Raster Analytics Conceptual Overview

ArcGIS

Enterprise

Web GIS Layers

GIS Data & Imagery

import and optimize

(optional)

New Web GIS Layers

distributed raster analytics cluster

ArcGIS Pro

(Users, Analysts, Researchers)

Design & Run Model

Model Execution Distribution

Developers & System Integration

Raster Analytics can power systems that need to

execute spatial analysis and image processing

models in a distributed and scalable environment.

It is designed for users, developers, and system

integrators.

Results are stored in distributed

storage and are immediately

available as new Web GIS Layers

which are already optimized for

further analytics

distributed raster datastore

GdbFilesWCS ServicesArcGIS Services

analysis results as a new Web GIS Layers

Web UX

ArcGIS 10.7Raster Analytics Test Cases

Raster Analytics Test Case: Terrain Suitability

Global SRTM 90m

0

100

200

300

400

500

600

700

800

1 2 4 8 16

790

425

252

126

80

Min

ute

s

Raster Analytics Processorsesri virtual machine

• 16GB RAM, 8 cores, NAS storage

13.12 hours

80 minutes

terrain suitability model• compute slope

• compute aspect

• remap

• overlay

global terrain suitability raster

Raster Analytics Test Case: Landsat Processing

(foreach) input scene

top of atmosphere

correction

modified soil adjusted

vegetation index

remap to classes

mask

no data

output thematic raster

Distributed Raster Analytics Cluster

• single node

• AWS c3.8xlarge

• 60GB RAM, 32 cores, 500GB SSD

• 200 Raster Analytics Processors

Infrastructure ProcessingInput Collection Output

Landsat GLS 1990

• 7422 Multispectral Scenes

Thematic Rasters

• 7422 Thematic Rasters

• Distributed Raster Datastore

2 hours 48 minutes44 scenes per minute

¾ scene per second

Thank you!