Table 1 Computer vision-based navigation and wayfinding systems
References | Beneficiary | Computer-vision solution | Path planning solution | Remarks /findings |
|---|---|---|---|---|
Lie et al. [29] | People with VI | Google Tango VPS | Prism MST algorithm, A* algorithm | (+) Haptic feedback system provided safe navigation in noisy environments |
Tian et al. [32] | People with VI | Canny edge detector, Tesseract and Omni page OCRs | Not available | (−) Path planning module is absent |
Lee and Medioni [33] | People with VI | Corner-based motion estimator algorithm | SLAM and D* Lite algorithm | (−) Inconsistency in constructed maps |
Garcia and Nahapetian [37] | People with VI | Canny edge detector and Hough line transform | Not available | (−) Detection failed for bulletin boards as well as low contrast wall pixels |
Manlises et al. [38] | People with VI | Image subtraction, Histogram backpropagation | D* algorithm | (−) Low brightness and noise in indoor areas will affect the recognition and feedback systems, respectively |
Bai et al. [39] | People with VI | Deep learning-based object recognition, scene parsing, Currency recognition functions | Vision-based slam | (+) Improved location awareness for the users |
Athira et al. [49] | Customers of shopping mall | Gist descriptors | Not available | (−) Does not support navigation between floors |
Pearson et al. [50] | Visitors of library | Bar code recognition | A* algorithm | (−) Misplaced books and books without barcodes can limit the system functionalities |
Li et al. [51] | Normal people | SIFT descriptors | Self-adaptive dynamic-Bayesian network | (+) Scalability |