Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Real-time Traffic Sign Detection and Classification: Evaluation of Image Processing performed on an FPGA-based platform
KTH, School of Industrial Engineering and Management (ITM), Machine Design (Dept.), Mechatronics.
2018 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

As a part of the development of autonomous vehicles and advanced driver-assistance systems (ADAS), vision systems are used as a method for collection of extensive data about the surrounding world [1]. This data can thereafter be processed and information can be extracted. Due to the safety-critical nature of automotive applications, the image processing of the camera stream must be performed in real-time [2].

This thesis investigates how a system with real-time performance potential - an FPGA-based system - can be utilised to perform image processing applications. Specifically the thesis looks into the research area of traffic sign detection and classification. A literature study is presented, along with a proposed implementation of a Traffic Sign Detection and Classification (TSDC) system.

The conclusion of the literature study is that many different methods have been tested previously but their performances are hard to compare. One of the most common approaches for FPGA-based implementations was chosen, due to its simplicity yet proven high accuracy by previous FPGA-based implementations. The approach - a colour thresholding and template matching - was partly implemented using the manufacturer Xilinx’s developing tool Vivado and High Level Synthesis (HLS).

The entire system was never implemented due to lack of time. However, the colour thresholding part of the algorithm was implemented and provided good result with a throughput of 209 frames/s, very low hardware utilisation and a low power consumption of 0.016 J/frame. This was determined using Vivado’s Design Evalu-ation tools.

A proof of concept was provided for the classification part of the system, that was never implemented on the platform, which showed that the classification part would likely constitute a performance bottleneck to the system.

The detection and classification results proved that if there was a sign in the image it was found 96.0 % of all cases on previously unseen data, but of those where only 79.0 % classified as true positives. In addition to this 34.9 % of the previously unseen images not containing the searched-for sign, were a false positive.

The conclusion of the thesis is that for a full system to be implemented, more of the tasks need to be performed on the FPGA, in order to have the potential to perform in real-time. One proposal to achieve this, is to implement a region of interest extraction, so only a single scale template match could be performed.

However, given the classification results, it is probably a too simple classifier for the problem. Another conclu-sion is therefore that a more sophisticated classifier would be of interest to test instead.

Abstract [sv]

Vision system är idag en ofta förekommande metod för informationsinhämtning för autonoma fordon och Advanced Drivers Assistance Systems (ADAS) [1]. Paå grund av dessa applikationers säkerhetskritiska natur måste den efterkommande bildbehandlingen ske i realtid [2].

Detta examensarbete undersöker hur en plattform med realtidspotential – ett FPGA-baserat system – kan utnyttjas för att utföra bildbehandlingsapplikationer. Mer specifikt undersöker detta arbete forskningsområdet att detektera och klassificera trafikskyltar. Initialt presenteras en litteraturstudie and därefter en implementation av ett Traffic Sign Detection and Classification (TSDC) system.

Slutsatsen som kan dras av litteraturstudien är att många olika metoder har tidigare testats, men att dessa ¨ar svåra att jämföra mot varandra. En av de vanligaste metoderna bland FPGA-baserade implementationer valdes, pga. dess enkelhet samtidigt som den visat hög precision vid tidigare FPGA-baserade implementationer. Metoden -en nyansbaserad färgseparering i kombination med figurmallmatching - implementerades delvis med hjälp av tillverkaren Xilinxs utvecklingsverktyg Vivado och High Level Synthesis (HLS).

Det fullständiga systemet implementerades aldrig pga. tidsbrist, men färgsepareringsdelen av algoritmen implementerades och visade på gott resultat med en kapacitet på° 209 bilder/s, ett lågt hårdvaruutnyttjande och en energiåtgång på° 0,016 J/bild, vilket kunde fastställas med hjälp av Vivados Design Evaluation verktyg.

Ett konceptbevis är försett för klassificeringsdelen av systemet, som aldrig implementerades på plattformen, vilket antydde att klassificeringsdelen av systemet rimligen skulle utgöra en flaskhals för prestandan.

Detektering- och klassificeringsresultatet tydde på att om den sökta skylten fanns i en bild så detekterades dess position 96.0 % av fallen på tidigare osedd data, men endast 79.0 % av dem klassificerades som true positives. Dessutom i 34.9 % av fallen, när en bild inte innehöll den sökta skylten, resulterade i en false positives.

Slutsatsen av det praktiska arbetet är att för att ett fulländat system ska kunna implementeras, måste mer av algoritmen laggas på FPGAn, för att kunna uppnå realtidspotential. Ett förslag för att uppnå detta är att implementera en identifiering av potentiella områden, där det kan finnas en stoppskylt. Detta skulle innebära att processorsystemet endast skulle behöva utföra figurmallmatching på en skala, istället för på 16 olika möjliga skalor. Emellertid är det möjligen inte värt besväret, då klassificeringsresultatet inte borde anses gott nog, vilket gör att en mer sofistikerad klassificeringsmetod vore önskvärd.

Place, publisher, year, edition, pages
2018. , p. 126
Series
TRITA-ITM-EX 2018 ; 691
National Category
Mechanical Engineering
Identifiers
URN: urn:nbn:se:kth:diva-244397OAI: oai:DiVA.org:kth-244397DiVA, id: diva2:1290265
Educational program
Master of Science in Engineering - Design and Product Realisation
Supervisors
Examiners
Available from: 2019-02-20 Created: 2019-02-20 Last updated: 2019-02-20Bibliographically approved

Open Access in DiVA

fulltext(26171 kB)291 downloads
File information
File name FULLTEXT01.pdfFile size 26171 kBChecksum SHA-512
58271524f38bb0078fda0b4ffdd125f084817397fc1e48fb36e55889373e7ed2090f8fdb44bbd7e4065124cd2e79e5678d9ebd60d1dbabb431238f8e4283eb73
Type fulltextMimetype application/pdf

By organisation
Mechatronics
Mechanical Engineering

Search outside of DiVA

GoogleGoogle Scholar
Total: 291 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 235 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf