Top Banner
Study of Sign Characteristic s on I-75, I- 85, and I-20 Morris Smith CEE 2699 Spring 2014 Dr. Yichang Tsai
18
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Smith Morris - Sign Characteristics [FINAL]

Study of Sign Characteristics on I-75, I-85, and I-20Morris Smith

CEE 2699

Spring 2014

Dr. Yichang Tsai

Page 2: Smith Morris - Sign Characteristics [FINAL]

Contents1. Introduction

i. Importance of Signsii. Sign Management

2. Objective3. Ground Truth Establishment

i. Methodology for Sign Data Extraction

ii. Methodology for Sign Data QA/QC

4. Ground Truth Summaryi. Statisticsii. Observations

5. LiDAR-Based Sign Detactioni. Methodologyii. Observationsiii. Limitations

Page 3: Smith Morris - Sign Characteristics [FINAL]

Introduction – Importance of Signs

• Traffic regulation and guidance• Symbols that are simple and easy to comprehend

• Safety• Especially important on freeways and highways

• Infrastructure development• Travel efficiency increased

Page 4: Smith Morris - Sign Characteristics [FINAL]

Introduction – Sign Management

• Traffic sign databases must be updated on a regular basis• Understanding several correlations

involving sign type and location

• Provides information to determine sign placement/replacement

• Advances in technology allow for quicker and broader analysis

Page 5: Smith Morris - Sign Characteristics [FINAL]

Objective

• To assess and improve the effectiveness of using LiDAR technology for sign detection• Establish a ground truth as a control for experimentation

• Fine-tune key parameters in LiDAR-based sign detection and record observations

• Based on observations, make recommendations to improve the current LiDAR-based sign detection method

Page 6: Smith Morris - Sign Characteristics [FINAL]

Ground Truth Establishment – Methodology for Sign Data Extraction

• Inventory signs along three major interstate highways in Georgia using Trimble Trident Analyst for Spatial Imaging software to designate GPS coordinates• Construction signs, overhead signs, informatory signs, mile

markers, cautionary signs, street signs• NOT inventoried: Temporary signs, guardrail markers, electronic

signs, signs on right-of-way on ramp

• Morris: I-20 East and West• Alabama – Atlanta and Atlanta – South Carolina

Page 7: Smith Morris - Sign Characteristics [FINAL]

Types of Signs Inventoried on I-20

Page 8: Smith Morris - Sign Characteristics [FINAL]

Ground Truth Summary – Statistics

• I-20• Total number of signs inventoried: 3,847

• Total number of frames: 136,016West of Atlanta

East of Atlanta

Eastbound 631 1,251

Westbound 589 1,371

Total 1,220 2,622

Number of Signs Inventoried on I-20

Page 9: Smith Morris - Sign Characteristics [FINAL]

Ground Truth Establishment – Methodology for Sign Data QA/QC

• Perform cross-checks with other researchers who extracted data on other highways• Ensures consistency among datasets

• Morris – reviewed Yutong’s data on I-75 North and South• Florida – Atlanta and Atlanta – Tennessee

• Observations• Dataset is considerably larger than that of I-20

• Many temporary signs that were erected during construction were inventoried

Page 10: Smith Morris - Sign Characteristics [FINAL]

81,288 frames

QA/QC – Observations

Page 11: Smith Morris - Sign Characteristics [FINAL]

Ground Truth Summary – Observations

• Overpass street signs were difficult to accurately inventory

• Sign markers for the highway in one direction appeared inaccurate when travelling in opposite direction

• Some signs were blocked by large passing vehicles – these coordinates are approximated

Page 12: Smith Morris - Sign Characteristics [FINAL]

Ground Truth Summary – Observations(2)

Page 13: Smith Morris - Sign Characteristics [FINAL]

LiDAR-Based Sign Detection

• Experimented with given parameters and ran LiDAR detection tool

• Compared results (sign location, number of signs inventoried, etc.) with collected ground truth data

• Based on comparison results, parameters were refined to achieve the best set for optimal sign inventory performance

Page 14: Smith Morris - Sign Characteristics [FINAL]

LiDAR-Based Sign Detection – Methodology

• The only parameters that were considered were the first five in the given window• The other parameters

were left alone

• Compared results via visual analysis with ground truth and other sets

• Used I-20 Eastbound, West of Atlanta to test results

Page 15: Smith Morris - Sign Characteristics [FINAL]

LiDAR-Based Sign Detection – Observations

Sensitivity 0.70

Minimum Elevation 2 ft.

Min. Dist. Sign to Sign

12 ft.

Max Lateral Distance

100 ft.

Minimum Hit Count 10

Number of Signs 408 signs

• Many signs that had multiple detections in previous dataset had only one marker in this set

• Used initially established dataset as control on which to improve

• Picked up a high percentage of ground signs

• Many signs were detected multiple times, e.g. contained multiple markers

*only the vertical camera playback was entered when testing this dataset

Sensitivity 0.70

Minimum Elevation 2 ft.

Min. Dist. Sign to Sign

12 ft.

Max Lateral Distance

100 ft.

Minimum Hit Count 15

Number of Signs 465 signs

Page 16: Smith Morris - Sign Characteristics [FINAL]

LiDAR-Based Sign Detection – Observations(2)

Page 17: Smith Morris - Sign Characteristics [FINAL]

LiDAR-Based Sign Detection – Limitations

• Limited time to test parameters

• Many ground truth markers were not completely accurate in locating the center of the sign

• Sign locations with more than one sign (i.e. I-20 signs with direction sign) were technically counted as one sign with LiDAR detection

• Some stationary trucks were detected by LiDAR cameras

• LiDAR only inventoried signs on the right-hand camera

Page 18: Smith Morris - Sign Characteristics [FINAL]

Questions?