Research And Application Of Automatic Palletizing For Cigarette Sorting Robot Based On Binocular Vision

The application of industrial robots is becoming more and more widely.And the automatic palletizing system based on binocular vision algorithms plays a important role in theory and practical practice.The main research content and achievements of this thesis are as follows:Firstly,the common recognition and localization algorithms of machine vision at home and abroad are compared and analyzed.According to the requirements of industry and characteristics of cigarette package,an identification and positioning system for irregular cigarette packages is designed.The charge coupled device(CCD)camera and lens of the system are selected.Secondly,an improved cigarette package identification method is proposed in this thesis based on the requirements of the irregular cigarette package identification and positioning system.This method firstly performs edge detection on the preprocessed image to obtain the edge image of the cigarette package.Then,the number of irregular cigarette packages are determined by further processing the edge image.An improved Canny algorithm is proposed in this thesis,in order to improve the problem which unfavorable edge detection caused by the interference of the heat shrinkable film on the surface of the cigarette pack and the diversity of the irregular cigarette package.A four-direction Sobel algorithm is proposed in improved Canny algorithm,aimed at solving the problem that the two-direction Sobel detection in Canny edge detection is not ideal.At the same time,for the problem that the fixed threshold in the Canny algorithm may generate false edges or more incidental noise,a gray histogram iteration method is proposed to determine the optimal high and low thresholds.The experiments show that the identification algorithm of this thesis can better complete the number of irregular cigarette package identification,and the average recognition time is about 0.57 s..Subsequently,the binocular vision is calibrated to obtain internal and external parameters.And the acquired image is corrected using the obtained parameters.in the positioning of irregular cigarette packages,an improved FAST-SURF algorithm is proposed to solve the problem that long time is taken and low accuracy is taken effect in existing algorithm cigarette positioning.The 16 contrast pixels of the FAST algorithm are recuced to 12 to improve the detection speed in improved FAST-SURF algorithm.TheFLANN algorithm is used to quickly match the feature vectors.The RANSAC algorithm is used to remove the mismatched pairs in the image.The experiments indicate that the matching time of the algorithm is about 0.64 s,and the average positioning error is about8 mm.The improved algorithm effectively solves the problem.Finally,an irregular cigarette package identification and positioning system was constructed,and the hardware and software design of the system was completed.The simulation algorithm was used to verify the algorithm in the system.The average recognition time of the experiment is 0.56 s,and the average positioning time is 0.54 s.The positioning error,identification and positioning time meet the system requirements.The experiment proves that the system can identify the irregular cigarette packet model and can accurately locate it.The experimental results show that the proposed method for identifying and locating irregular cigarette packages with improved Canny algorithm and improved FAST-SURF algorithm is effective and feasible.