Study On Image Processing Technology For A Mini-AGV Under A Type Of Incandescent Illuminant

Automated Guided Vehicle is a representative device in Flexible Manufacture System, which can be assembled in any manufacturing system. Meanwhile, it is very easy to transfer parking location and moving path. With improvements of technology, AGVS is going to be adopted by almost all the manufacturing systems in future to result in a great transform in unmanned-transportation. And its usage means will be different accordingly.AGV comprises many technologies such as sensor, signal processing, computer, automatic control, communications and mechanics. With advantages of flexible and efficient transport, energy saving, reliable running, and little pollution, AGV has appeared in many fields as supermarkets, workshops, large-scare automatic warehouses, hospitals and purvey centers. In a word, it is one of the most primary symbols in automatic industries.AGV can be classified into different types according to different navigating means. Among them, the magnetic inductive guiding method is mainly used for most commercial AGVs. Although this technology is quite mature, it has some shortcomings,such as high cost and long period of long navigation path building and maintenance, difficult modification. As a result, it lacks navigation flexibility to some extent. Further more, this type of AGV cannot perform well when there is serious electromagnetic disturbance.AGV which uses the navigation of computer vision, provides a large of information so as to conquer above disadvantage. Group always works hard in the field and acquires some success. According to actual power and Technology in China, the difficult technically AGV is expensive and infrequent, furthermore, it will also have good foreground.In order to reducing manufacture cost, a new type of AGV (JLUIV-3C) is established and has designed illumination by myself. The illuminations confirm the text's title, and that the text tries to study on image processing in complicated condition.Title of the dissertation is Study on Image Processing Technology for a Mini-AGV under a type of Incandescent Illuminant, and study on mainly image processing. The paper is made up of four parts: 1. The design of the computer vision system and the equalization of illumination. 2. The navigation using lane mark. 3. The recognition of Numeric numbers of site where load and unload work pieces. 4. The experiments of the algorithms'reliability and applicability.The computer vision system of JLUIV-3C is made up of software and hardware. Illumination of the camera is also devised. After expatiation of the working principles of the vision system, the factors which can affect the feature extraction are introduced in detail. As illumination isn't smooth, the paper bring forward a equalization way of illumination based on Gamma principle. The path recognition is a key technology in the navigation of AGV based on computer vision, which should have some abilities such as higher reliability and precision. The paper makes use of fast Otsu method to segment an image dynamically and edge enhancement arithmetic to extract edges. The paper compounds Hough transformation and improved line scanning to locate the moving path. And accurate identifications are derived easily in complicated cases. Meanwhile, I design a method of eliminating noisy points which is proved to be effective in experiments.Number recognition in workplaces is an advantage of AGVs based on computer vision and it is a difficult part also. The paper takes the first lead to use colorized numbers that provide a large number of information. The segmentation of numbers integrates HSV space and gray space, which assures less noises and integrality of number as much as possible. Both region growth method and removing isolated points method are also used to process an image, and the horizontal and vertical projection is used to confirm numbers'AOI. In the end, template method and LED number recognition based on rotated fixity are adopted to figure out the number. The offset angle of number decides which of them will be adopted. The template method can resist bad pollution and recognize incomplete figure, while the other method may avoid some outside disturbances.The software application system is developed with Visual C++. Fuctions of the algorithms proved to be reliable and robust. The experimental results prove the validity of the algorithms mentioned in this paper. Firstly, nine templates are used to perform illumination equalization including removing lamp spot and whole image equalization. The data disposed with Matlab prove the validity of the method. At the same time, segmented images of paths are compared after equaling illumination. Secondly, path recognition outcomes prove the reliability of the algorithms mentioned above with some templates.As for the number recognition in workplaces, a lot of experiments ensure the validity and reliability of the two methods in even complicated circumstances.