Research On Indoor Obstacle Avoidance Method For Visually Impaired Based On Grid Laser

For blind people walking obstacle avoidance,accurate and real-time obstacle detection and information tips is the core of its research.Identify information about the surroundings of the blind person during walking and give instructions accordingly,with the aim of preventing them from colliding with obstacles to prevent injury.There are many ways to detect obstacles,such as the use of ultrasound,infrared and other types of sensors.However,since ultrasonic detection can only obtain distance information,it does not fully reflect the information of the environment around the blind.In addition,due to the diversity of obstacle types and shapes,data acquisition equipment and detection algorithms will be affected by the environment.Global Positioning System(GPS)can be used for blind walking avoidance,but indoor GPS the positioning system will lead to uncertainty due to environmental impact.Therefore,how to improve the accuracy of obstacle detection has become a hot topic in the current research,in order to improve the safety of blind walking.This paper mainly studies the obstacle detection based on grid laser,taking into account the interference factors such as different shape of obstacles,uneven light and ground background in indoor environment.The main research contents of this paper are as follows:(1)Aiming at the problems in the detection of existing obstacles,this paper chooses to use red grid laser and camera combination equipment to study the obstacle detection algorithm.First select the appropriate laser equipment,then use the selected equipment for image data or video data acquisition,and finally set up an obstacle detection database.(2)Aiming at the characteristics of database image acquisition,an image preprocessing method based on color feature is used in this paper.First RGB converts the Lab color space,and then extracts a-channel image.Experiments show that this method can effectively remove the influence of light and ground background on laser.(3)According to the characteristics of grid lasers,a linear detection algorithm based on improved Hough Transform is proposed in this paper.Firstly,the image is preprocessed and binarized,and then the straight line detection results are obtained by the Hough Transform,and finally the results of grid laser line detection are obtained by sorting the data of line detection,calculating the grid line coordinates and drawing the grid line.Experiments show that the proposed algorithm is not only better than using Hough Transform directly,but also can detect grid laser line clearly and accurately,and can solve the problem of light shielding the laser lines.(4)In this paper,the geometric features of the image are detected by using Blob feature extraction grid line.Firstly,the Blob feature parameters are initialized and the coordinates and size of grid connected domain are initialized by using Blob features,then the threshold of obstacle detection is determined by experiment and the threshold value of grid connected domain coordinates and size is updated,and the detection results are obtained by threshold judgment.Experiments show that this method can achieve good results in obstacle detection.(5)According to the process of obstacle detection based on grid laser,this paper realizes the function of obstacle detection on the PC side,and the detection results are output with simple voice prompts.