Research On Visual Guidance And Safety Information Detection And Recognition For Visually Impaired People

In recent years,with China's aging population continues to increase,visually impaired people have increased dramatically,how to solve the security of the visually impaired people travels at the same time,improve the quality of life of the community,has attracted many attention of the society.Electronic guide system is a multidisciplinary cross discipline direction,involving machinery,electronics,communications,sensor,information processing,multiple categories such as the information processing is the key to embodying its intelligent degree.Because visually impaired people lack visual perception of surrounding environment information,and the diversity and complexity of walking environment,it becomes the bottleneck problem of information processing in electronic guide system.Aiming at this problem for visually impaired people oriented visual guide safety information detection and recognition research,concrete analysis,research and complete guide blind detection in blind system identification,the blind area in the obstacle detection and obstacle collision avoidance warning blind method in the existing problems and the insufficiency,in order to further enhance the intelligent degree of the electronic guide and security,provide convenience and safety for the visually impaired people travel assistance systems.The main research contents of this paper including:1.Identification and detection of blind road.Based on the visual guide system,as the blind road in the urban road is vulnerable to the interference of light,shadow and so on environment,based on the analysis of the blind road,on the basis of color features is proposed based on color feature fusion blind identification algorithm of the edge information.The algorithm through the analysis of the characteristics of yellow of the blind to the characteristics of multiple color space model and adopt accord with human visual perception Lab space model to extract b component,and the most between-cluster variance method of image segmentation,effectively segment the blind area.According to the edge features of the blind channel,Canny operator is adopted to extract the edge points of the blind channel,and the blind channel edge points are optimized by Hough detection,and the edge boundary of the blind channel is obtained.Experimental results show that the algorithm can effectively identify the blind region and reduce the influence of interference.2.Detection of blind area obstacles.Obstacle detection accuracy is related to the safety of the visually impaired people travels,due to the complexity of blind road along the route environment,and the presence of the obstacle with unpredictability and uncertainty,this paper is divided into right in front of the obstacles in the blind area and blind road on both sides of the suddenly appear obstacles.Front for the obstacles in this paper,we use single-frame image processing to create a region of interest for the acquired image to reduce the obstacle search range,and then perform edge detection in both directions for the region to further reduce the obstacle area.Finally,calculate the image entropy to determine if there is an obstacle in the area;the blind road on both sides of a sudden the movement of the obstacles in this paper,based on image sequence processing,based on the improved frame difference method fusion gray-scale projection algorithm to test the movement obstacle,this method can be used to detect the blind to the case of background motion suddenly appeared on both sides of mobile obstacles.3.Visual guide blind warning anti-collision model.In view of the problem of anti-collision warning for visually impaired people,this paper puts forward a visual guide blind anti-collision warning model based on the detection and localization of obstacles in the blind area.Under the hypothesis that the blind road pavement is horizontal,the world was determined by the imaging device coordinate system and image coordinate system of geometric relations,based on single objective visual range model,estimate the visually impaired and the actual distance between the obstacles.At the same time,the security distance is set up and a reasonable anti-collision warning strategy is set up to ensure the safety of visually impaired people effectively.4.Finally,Finally,a visual guidance blind security information detection and identification system is designed to verify the effectiveness of the proposed algorithm.Detailed hardware design and software flow charts are given.Through the video images obtained in different blind-path environments,the algorithm can identify obstacles and give corresponding strategies and comprehensive experimental results,verifying the overall performance of the proposed algorithm.