Abstract
In this research, we present a system that enables independent walking in individuals with visual impairment by recognizing tactile paving. The system warns the user about dangers posed by deviations from the walking path or possible collisions. Therefore, the tactile paving itself and obstacles on it must be detected. This paper discusses problems related to detecting tactile paving. To address these problems, a new camera-based method is introduced, which detects tactile paving automatically. The method relies on a dynamic threshold approach in the HSV color space, making the image processing more robust against varying lighting conditions, environments, and differences in color. Such an approach is not possible using the methods based on fixed thresholds reported in the literature. The results of our experiments confirm that high rates of tactile paving detection were achieved under various conditions.
Highlights
According to World Health Organization statistics (World Health Organization, 2019), as of 2019, there are 2.2 billion people around the world who have some form of visual impairment or blindness [1]
In 2015, 216.6 million people had moderate to severe visual impairment, and this is estimated to rise to 550 million people in 2050 [2]
This paper makes the following contributions: 1) proposed algorithm obtained a high accuracy of 91.65% when it was tested with worldwide tactile pavement images and images without tactile pavements, which represents a detection accuracy higher than that of existing methods in the literature, 2) the proposed algorithm obtained a detection rate of 93.36% when it was tested with the same worldwide tactile images, 3) the proposed method is robust because it can detect indoor/outdoor tactile paving with different colors, under the different lighting conditions
Summary
According to World Health Organization statistics (World Health Organization, 2019), as of 2019, there are 2.2 billion people around the world who have some form of visual impairment or blindness [1]. Some walking support systems target the protection of visually impaired people from collisions by recognizing only obstacles that are present during the walk. These systems are developed by using sensing devices such as stereo cameras and ultrasonic sensors. Support systems for individuals with visual impairment can take advantage of this infrastructure by automatically detecting tactile paving and obstacles and providing corresponding guidance. Based on the HSV histograms, dynamic thresholds are statistically determined These thresholds are applied to detect tactile paving region from the images. This paper makes the following contributions: 1) proposed algorithm obtained a high accuracy of 91.65% when it was tested with worldwide tactile pavement images and images without tactile pavements, which represents a detection accuracy higher than that of existing methods in the literature, 2) the proposed algorithm obtained a detection rate of 93.36% when it was tested with the same worldwide tactile images, 3) the proposed method is robust because it can detect indoor/outdoor tactile paving with different colors, under the different lighting conditions
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have