Automatic Bubble Level Correct System Based On Deep Learning

With the development of digital image processing,machine vision has been widely used in industrial production.Deep learning has become a new hot research field which makes intelligent detection possible.This paper focuses on automatic bubble level correct system to correct the bubble level precisely.In this paper,we first introduce the background of the topic,the research status and development trend of the algorithms of edge detection and deep learning object detection.Then we introduce the basic algorithms of Convolutional Neural Network and the core ideas of the object detections model.Aiming at the automatic correction of the bubble level with large angle of inclination,we collect and label bubble level images with different light conditions to build data set.We analyze the effect of different anchor numbers on YOLO model's performance by adopting K-means clustering algorithm.Faster R-CNN,YOLO and SSD are three of the best object detection models.Three models' detection accuracies and average IOU were compared.We analyze the edge detection method based on mathematical morphology and use Zhang parallel algorithm for thinning to extract single pixel edge.After that we design two single pixel edge fitting methods combining Progressive Probabilistic Hough transform with least square method.We compare the two methods' accuracies on the single pixel edge extracted by mathematical morphology gradient with Zhang parallel thinning and Canny edge detection.At last we introduce the structure of automatic bubble level correct system.Aiming at this application,we use Client/Server network structure and design the software.The images were sent to the server and calculated.The calculation results were sent back to the client to control motor to correct the bubble level.