The Research And Design Of Portable Guide System

According to the statistics released by the World Health Organization in October,2018,it is estimated that there are about 1.3 billion people suffering from different forms and degrees of vision impairment globally,and most of them are over 50 years old.In recent years,with the development of computer software and hardware electronic equipment,the development and application of micro computer system has been widely concerned.Many portable electronic devices have emerged to enrich people's lives,which also makes people more capable to help the vulnerable group of blind people.The existing intelligent blind assistant tools can help people with vision impairment to some extent,but it is difficult to be popularized because of its high price and poor user experience.In view of the above problems,this paper designs a portable and intelligent guide system to help blind people avoid obstacles in the journey of moving.The system captures the video images in front of user by an installed camera,analyzes and processes them using the image processing technology,and reminds the user on voice.Firstly,this paper proposes an obstacle detection method based on the mean differences between classes to calculate the contour shape of obstacles.This method is based on the saturation characteristic which keeps good illumination robustness,takes the neighborhood of histogram valley as the constraint condition to balance the difference on sizes,finds the best threshold to meet the basic Otsu criterion and maximizes the difference between the mean values of classes and the overall mean value.After that,a monocular vision ranging model based on perspective transformation is established,and the mapping relationship among the world coordinate system,camera coordinate system,imaging plane coordinate system and pixel coordinate system is derived step by step through the coordinate transformation of geometry space;the camera is calibrated by the chessboard method of Zhang Zhengyou,so as to obtain the internal and external parameters of the camera and lens distortion factor;the relative position information of the obstacle target is calculated.Finally,the above obstacle detection and ranging algorithms are integrated into raspbian based raspberry pi to complete the overall design of the blind guide system.The system uses raspberry Pi 2b as the development platform,after connecting the hardware part,uses Python and OpenCV library interface to integrate and transplant the above algorithms;The tests are carried out,and the experimental results prove the functionality and stability of the system.