The Design Of Bamboo Detection System And Positive And Negative Of Bamboo Research Of Recognition Algorithm

Bamboo mat products have became the most important necessity goods in the summer. Especially, the mahjong bamboo mats, which make people feel cooler, become more and more popular. With it playing a role in the market, how to control its qualities and improve its production has attracted producers’ attention. In fact, among10product processes, only the bamboo pieces selection need artificial selection and the rest have been implemented by machines. Therefore, the production efficiency of bamboo pieces selection fall behind the rest processes. What’s more, different optical responses of workers to bamboo pieces make it difficult to meet the high requirement to quality. To solve the above problems, the papers design a bamboo pieces selection device and present the recognition algorithm to take the place of artificial selection. Besides a high quality, they also have a high efficiency and will be popular in the future.The bamboo pieces selection device is built in the two-layer architecture whose each layer is same, including mechanical device, hardware circuit and control software.And the mechanical device includes vibration feeding device, transmission device, poses adjustment device, image acquisition device and turn-over device. The vibration feeding device consists of electromagnetic vibration and pathway. It can finish the feeding process by electromagnetic vibration to make bamboo pieces in a sequence along the spiral pathway. The transmission device is made up of conveyor belt, conveyor belts wheel and driver, driven wheel. After calculating the need transmission ratio to get the suitable speed of bamboo pieces transmission, it can implement the function of transmission.ThenIn order to adjust the position of bamboo on the conveyor belt; we design the device to adjust the pose of machine by the mechanical device’s guidance quality. And image acquisition device install industrial camera and ring illuminator to take pictures of bamboo. Turn-over device drive the bamboo down along the inwall of curved pipe with the gravity’s help to turn the bamboo over.The hardware circuits include driver module, host system, infrared detection module, and communication module. As the main center to link all modules, the host system is based on AT89S52.The driver module get charge of24V circuit by high and low level to handle the dynatron’s open or close. Infrared detection modules finish the bamboo’s existence detection by RPR220infrared probe, LM393comparator and two-stage inverter. Communication modules conduct the communication connection between upper computer and MCU.The control software is compiled in C and finished by keil4.Host procedure; Communication procedure and checking routine develop the control software. And host procedure call upon the rest, communication produce finish receiving and transmitting data by interrupt.The device presented by the paper can finish the whole process without manual work.Besides the low cost, high recognize rate and good stability, it also improve the selection efficiency. After a lot of compiles and tests, the device can run perfectly with5pieces every second and lay the root of bamboo selection.Due to the bamboo selection need to select both from the positive and the negative, it’s necessary to recognize the positive and the negative in order to improve the rate of bamboo detection. The papers present an algorithm to recognize the positive and the negative. The algorithm extract2textural features and gray feature build and train a3-layers BP network with3inputs and1output. The extracted feature is used to be input, through the network, will become the positive or negative predicted value. According to the predicted value, it can judge the positive or negative. The positive recognize rate reach98.42%,the negative reach99.28%.The high recognize rates both meet product’s requirement.To sum up, the circuit section and the mechanical part in the bamboo detection device designed in this paper can combine very well; the entire device can be stable running at the rate of5pieces per second. The proposed recognition algorithm is fast and steady to detect the positive and negative of the bamboo piece.Besides, it can run on the upper machine, such as Labview software. And the final recognition rate meets the actual requirements.