Research On Indoor Electronic Guide System Based On RGB-D Sensor

In recent years,RGB-D sensor is widely used in the field of indoor blind assisted walking due to its strong real-time and good synchronization.Based on the research of the existing RGB-D-based blind guidance system,it is found that these systems usually process the original depth image directly,neglecting the existence of holes in the depth image which reducing the accuracy of the system.Moreover,the system often needs human intervention in obstacle partitioning,and it has poor adaptability and reliability.Therefore,this paper aims at these two problems to improve the existing RGB-D indoor guidance system.In order to solve the problem of the hole in the depth image,an improved fast depth image hole restoration algorithm is proposed.Firstly,the algorithm aligns the acquired original color image and depth image.Then,according to the number of zero pixels in the adjacent area of the hole,the hole is divided into large hole and small hole.Finally,the improved pixel filtering method and the improved joint bilateral filtering algorithm are used to repair the two kinds of holes from small to large.Compared with the experimental results,the algorithm can ensure the accuracy of the image repair while having higher execution efficiency.In this paper,an edge-based automatic seed point depth image segmentation algorithm is proposed to solve the problem of artificial interference in the segmentation of obstacles.The algorithm finds the edge of the object in the image by the adaptive edge detection algorithm,and looks for the pixels on the edge that belong to the foreground.Finally,the foreground edge pixels are used to be seeds for regional growth.The experimental results show that the algorithm is superior to the traditional algorithm in selecting seed points,and the accuracy of the segmentation depth image is high.Finally,in order to solve the problem that the existing RGB-D electronic guidance systems neglects to repair the original data defects and needs manual intervention.Based on the proposed depth image restoration algorithm and the adaptive depth image segmentation algorithm,an indoor blind assistant system is designed.Through the detection of many sets of scene depth images,the system can find out the position of the obstacle in the scene and make the corresponding obstacle avoidance strategy according to the position of the obstacle.In the process of system execution,no additional manual processing is needed and it solves the deficiencies in the existing system.