Napredna pretraga

Pregled bibliografske jedinice broj: 928338

Real-time Multiresolution Crosswalk Detection with Walk Light Recognition for the Blind


Romić, Krešimir; Galić, Irena; Leventić, Hrvoje; Nenadić, Krešimir
Real-time Multiresolution Crosswalk Detection with Walk Light Recognition for the Blind // Advances in Electrical and Computer Engineering, 18 (2018), 1; 11-20 doi:10.4316/AECE.2018.01002 (međunarodna recenzija, članak, znanstveni)


Naslov
Real-time Multiresolution Crosswalk Detection with Walk Light Recognition for the Blind

Autori
Romić, Krešimir ; Galić, Irena ; Leventić, Hrvoje ; Nenadić, Krešimir

Izvornik
Advances in Electrical and Computer Engineering (1582-7445) 18 (2018), 1; 11-20

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
Assistive technology ; image recognition ; machine vision ; morphological operations ; object detection

Sažetak
Real-time image processing and object detection techniques have a great potential to be applied in digital assistive tools for the blind and visually impaired persons. In this paper, algorithm for crosswalk detection and walk light recognition is proposed with the main aim to help blind person when crossing the road. The proposed algorithm is optimized to work in real-time on portable devices using standard cameras. Images captured by camera are processed while person is moving and decision about detected crosswalk is provided as an output along with the information about walk light if one is present. Crosswalk detection method is based on multiresolution morphological image processing, while the walk light recognition is performed by proposed 6- stage algorithm. The main contributions of this paper are accurate crosswalk detection with small processing time due to multiresolution processing and the recognition of the walk lights covering only small amount of pixels in image. The experiment is conducted using images from video sequences captured in realistic situations on crossings. The results show 98.3% correct crosswalk detections and 89.5% correct walk lights recognition with average processing speed of about 16 frames per second.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo



POVEZANOST RADA


Ustanove
Fakultet elektrotehnike, računarstva i informacijskih tehnologija Osijek

Časopis indeksira:


  • Web of Science Core Collection (WoSCC)
    • Science Citation Index Expanded (SCI-EXP)
    • SCI-EXP, SSCI i/ili A&HCI
  • Scopus


Citati