Title: Recognition and localisation of staircases for visually impaired people

Authors: Jyoti Madake; Shripad Bhatlawande; Rutuja Nagdekar; Shweta Munjewar; Mitali Nimase; Rahul Paikrao

Addresses: Department of Electronics and Telecommunication Engineering, Vishwakarma Institute of Technology, Pune, 411037, Maharashtra, India ' Department of Electronics and Telecommunication Engineering, Vishwakarma Institute of Technology, Pune, 411037, Maharashtra, India ' Department of Electronics and Telecommunication Engineering, Vishwakarma Institute of Technology, Pune, 411037, Maharashtra, India ' Department of Electronics and Telecommunication Engineering, Vishwakarma Institute of Technology, Pune, 411037, Maharashtra, India ' Department of Electronics and Telecommunication Engineering, Vishwakarma Institute of Technology, Pune, 411037, Maharashtra, India ' Department of Electronics and Telecommunication Engineering, Vishwakarma Institute of Technology, Pune, 411037, Maharashtra, India

Abstract: Vision-based guidance to detect the staircase present in the scene and know its direction will be beneficial for blind people to avoid accidents. This paper presents a real-time embedded system implemented using Raspberry Pi embedded processing device with a portable camera and audio alert device. The paper summarises the literature presented for assisting blind people in staircase recognition. The proposed system performs feature extraction using an AKAZE detector and descriptor. The feature vector obtained is optimised using K-means clustering and principal component analysis. The optimised feature vector is trained using four different classifiers KNN, SVM, random forest, and AdaBoost. The performance of the classifiers is evaluated using accuracy, precision, F1 score, sensitivity, specificity and AUC metrics. The system reliably detects the position of the staircase in the scene with 90% accuracy and alerts the blind user using feedback through earphones.

Keywords: staircase detection; visually impaired people; AKAZE; K-means; principal component analysis; PCA; AdaBoost.

DOI: 10.1504/IJCVR.2025.149863

International Journal of Computational Vision and Robotics, 2025 Vol.15 No.6, pp.719 - 738

Received: 03 Apr 2023
Accepted: 15 Nov 2023
Published online: 14 Nov 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article