Research on the Method of Displaying the Contour Features of Image to the Visually Impaired on the Touch Screen

Conveying image information to the blind or visually impaired (BVI) is an important means to improve their quality of life. The touch screen devices used daily are the potential carriers for BVI to perceive image information through touch. However, touch screen devices also have the disadvantages of limited computing power and lack of rich tactile experience. In order to help BVI to access images conveniently through the touch screen, we built an image contour display system based on vibrotactile feedback. In this paper, an image smoothing algorithm based on convolutional neural network that can run quickly on the touch screen device is first used to preprocess the image to improve the effect of contour extraction. Then, based on the haptic physiological characteristics of human beings, this paper proposes a method of using the improved MH-Pen to guide the BVI to perceive image contour on the touch screen. This paper introduces the extraction and expression methods of image contours in detail, and compares and analyzes the effects of the subjects' perception of image contours in two haptic display modes through two types of user experiments. The experimental results show that the image smoothing algorithm is useful and necessary to help obtain the main contour of the image and to ensure the real-time display of the contour, and the contour expression method based on the motion direction guidance helps the subjects recognize the contour of the image more effectively.