An interpretation system for ducth cadastral system

22
Recognition system for building flop lines Poojith Jain-0666444

description

 

Transcript of An interpretation system for ducth cadastral system

Page 1: An interpretation system for ducth cadastral system

Recognition system for building flop lines

Poojith Jain-0666444

Page 2: An interpretation system for ducth cadastral system

Introduction

• Aim• Extraction of information from building flop lines• Storing extracted information• Representing information in CityGML

Page 3: An interpretation system for ducth cadastral system

Basic Concepts

• Graph•  Graph is an ordered pair G: =

(V,E) comprising a set V of  vertices together with a set E of edges.

• Graph is used to show connectivity of vertices.

• Computer Representation of images• Pixels• Pixel value based on the color• Array representation

Page 4: An interpretation system for ducth cadastral system

The Process

Flop line image of Building

Thresholding and Noise Removal

Labels Identification And image cleaning

Graph construction

CityGML Representation

Page 5: An interpretation system for ducth cadastral system

Building Flop Line

• Gray scale image• High Resolution• Indication

Thick lines ownership boundaryNumbers ownesrhip rightsLabels usage type

Page 6: An interpretation system for ducth cadastral system

Assumptions made

• Always thick lines indicates ownership boundary• Numbers always enclosed in a polygon• Single number in a polygon represents ownership• Numbers does not ovelap with lines and symbols

Page 7: An interpretation system for ducth cadastral system

The Process

Flop line image of Building

Thresholding and Noise Removal

Labels Identification And image cleaning

Graph construction

CityGML Representation

Page 8: An interpretation system for ducth cadastral system

Thresholding and Noise Removal

• Thresholding• Noise

• Gaps • Missing pixels

• Continuity is important for contour detection

• Solution• Closing Operation

Page 9: An interpretation system for ducth cadastral system

Closing Operation

CLOSING

Page 10: An interpretation system for ducth cadastral system

The Process

Flop line image of Building

Thresholding and Noise Removal

Labels Identification And image cleaning

Graph construction

CityGML Representation

Removing Labels and

Thin Lines

Number Identification

Page 11: An interpretation system for ducth cadastral system

Ownership Identification

• Identify the location of the labels• Connected component

labeling• Size criteria

• Extract the labels• Recognize the labels

OCR

OCR

{3,x,y}

{4,x,y}

Page 12: An interpretation system for ducth cadastral system

The Process

Flop line image of Building

Thresholding and Noise Removal

Labels Identification And image cleaning

Graph construction

Removing Labels and

Thin Lines

Number Identification

CityGML Representation

Page 13: An interpretation system for ducth cadastral system

Removing Labels and Thin Lines

• Labels indicate property usage and type

• Thin Lines indicate sub region information

• Thick lines indicate boundary

• Remove labels and thin lines.• Connected component

labeling• Opening operation

Page 14: An interpretation system for ducth cadastral system

The Process

Flop line image of Building

Thresholding and Noise Removal

Labels Identification And image cleaning

Graph construction

CityGML Representation

Corner Dection

Identifying Ownership boundary

Skeletonization

Page 15: An interpretation system for ducth cadastral system

Skeletonization

• Why Skeletonization?• Reduces foreground

regions in an image to a skeleton• By thinning operation

• Skeleton should be• One pixel width• Preserves connectivity• Preserves Topology• Centered

Page 16: An interpretation system for ducth cadastral system

The Process

Flop line image of Building

Thresholding and Noise Removal

Labels Identification And image cleaning

Graph construction

CityGML Representation

Corner Detection

Graph construction

Face and Floor

identification

Skeletonization

Page 17: An interpretation system for ducth cadastral system

Corner Detection

• Corenrs are intersection of two or more edges

• Corners forms the node of the graph

• Harris corner Detection• Invariant to

• Scaling• Image noise• Rotation• Illumination variance

Corner Detection

Page 18: An interpretation system for ducth cadastral system

Graph Construction

• Identify the nodes• Identify the edges• Optimization

Page 19: An interpretation system for ducth cadastral system

Graph Construction

• Identify the nodes• Identify the edges• Optimization

Page 20: An interpretation system for ducth cadastral system

The Process

Flop line image of Building

Thresholding and Noise Removal

Labels Identification And image cleaning

Graph construction

CityGML Representation

Corner Detection

Graph Construction

Face and Floor

Identification

Skeletonization

Page 21: An interpretation system for ducth cadastral system

Face Recognition

• Each enclosed face becomes ownership boundary

• Associate ownership• Store the information

{3,x,y}

{4,x,y}

3

4Ownership

Ownership Right

Point co-ordinate

Page 22: An interpretation system for ducth cadastral system

Floor Identification

• Identifying Floors• Storing Information

2 3 4 44

1 2 3 4

Ownership

Ownership Right

Floor Number

Co-ordninates

. . .