| The food is the material foundation that the mankinds rely for existence in the life activity; it is the mankind's vital energy. The relationship of food hygiene and the person's healthy are extremely closely. By the examination of the food microorganism, the food processing environment and the food hygiene circumstances can be judged, can to the food was been the evaluation of an exactitude by the polluting degree of germs, managing the work to provide science for various works according to, provide to infect and the prevention and cure measure that the food of the mankind, animal be poisoned. The germs total amount is the food quarantine medium the most important of a standard, the traditional development count the method currently, taking 48 hours to get an accurate result, have already canned satisfy less and less now increasingly and quickly living rhythm. Also released various fast examination method at home and abroad now, but all have thus or so of weakness, limited the expansion, can't replace development to count the position of the method completely. So this text puts forward to identify the technique according to the picture of the machine vision and the nerve network, used for fast examination of the germs total amount of the food.This paper designs before manufacturing handled the organization and analyzed the device automatically, handled to equip to used for before to leave the heart separation and dye germs, the automatic analysis equipped used for withdraw the germs picture combine as to it's characteristic analysis and count. The usage shoots to be like the head to take the sample germs picture, the usage picture operation, smooth etc. means carries on removing the background to the original picture, doing away with the garbage etc. the processing carries on the picture processing. The point carries on the characteristic to withdraw the work on this foundation. The standpoint that put forward to make use of the characteristics, such as the germs shape and color...etc. to carry on identify, and input the nerve network classification machine to carry on mode's identify, the end give examination result.According to the above-mentioned way of thinking, aim at the food to amount this research object in the germs, carry on the following work.1. Design the hardware device. Was divided into the automatic picture of the germs identify examine the system before to handle the organization and analyzed to equip two parts automatically. The certain front handles the adoption beautiful and blue dye the liquid, pass the auto is provided for a slice of equipments to transport toward microscope, to carry out high-efficiency movement of the system. In automatic analysis equip, the picture that chooses to use to take to photograph to connect oculars three WV- CP240 type colorful CCD of the eyes living creature microscope CAs XSP-8s(16×100), Panasonics to shoot to be like the head, 42BYG228 stepping electric motors and drivers,AT89S51 single chip, Shuangming SDK-2000 to collect the card and by oneself designs the development identifies and controls the software. The germs picture that guarantee to acquire possibly is clear; Was collect by colorful picture the card satisfy that equipments of the germs picture importation calculator can to the analysis of the sample germs total amount and count the function.2. The hardware system of pictures'automatism collection. The speediness examine of the germs in the food, the picture of the germs smear collected followed the principle is speediness, nicety, full-scale. According to this principle after the pretreatment, put the smear on the fixture of luggage carrier in order to make preparations for pictures'collection. In the collection of pictures, control the stepping electric motor by single chip. In order to drive the microscope's luggage carrier moving in the way of X-Y aspect. Pay equal attention to settle the choice of stepping electric motors and drivers. The removal rate, the fractionizing of drivers, and the system of single chip controlling the stepping electric motors.3. The software system of pictures'automatism collection. According to the resource of the hardware, we calculate the length of the X-Y direction axis'screw accurately. So we can get the impulse number when the stepping electric motors move one step. On the flat roof of VC++, on the basic of the assemble language using C language translate and edit by Keil uvision2 and transmitting to the CMOS chip of the single chip. So we can achieve the aim of controlling stepping electric motor's rotate speed, positive and negative running, and stepping instance act. This paper considers the application of single chip microcomputers in power supply control of step electric motors. It makes step motors run normally at an adjustable speed and in a subdividing running.4. Carry on preparing the processing to the picture and the picture partition. Because the picture adopts the concentrated interference garbage, and the light shine on one degree asymmetry etc. problem, affect hereafter partition the processed accuracy, past must make use of the picture operation, smooth etc. means carries on removing the background to the original picture, doing away with the garbage etc. processing. At the same time, aim at to partition to complete behind of picture have a lot of miscellaneous and loose garbage, also may exist the hair of the target outline to stab, glue to connect etc. phenomenon, make use of the appearance to learn of open, shut to match etc. the operation and artificial edit of method remove these not to measure the influence, to acquire good two values to turn the picture. The picture partition inquiring into various picture partitions the technique and it is in show minute details the picture partition of application, and applied various partition calculate ways to carry on the comparison to characteristics of show minute details the picture, choose to use finally again calculate way, make the accuracy that germs partition have the bigger exaltation.5. The characteristic choice. Carry on the choice to the characteristic that germs withdraw, totally chose to combine the success to withdraw the week is long, area, shape factor, rectangle degree, elongation degree, the characteristic ash an etc. index sign etc. six valid characteristic, selected by examinations a great deal of non- germs individual at the same time, withdraw their characteristic parameter, for identify the work to provide the foundation afterwards. 6. Identify of germs and count. Knot and this text the research object, six valid characteristics that aim at to withdraws, design also trained the BP network the classification machine, the establishment chooses to use to deliver the function adoption Sigmoid function, biggest training number of times is 1000 times, minimum training velocity is 0.9, the dynamic state parameter is 0.6, allow error margin is 0.0001, target of week long, area, shape factor, rectangle degree, the elongation degree and the characteristic ash degree, be the nerve network inputs the layer of node, output layer of node count to 2, use binary system of 1 and 0 mean, germs of tutor's signal is 1, non- germs individual of tutor's signal is 0, the node of the implicit layer count to 6, namely the network structure is 6-6-2.Carry on the training on the applied nerve network of the foundation of the comprehensive experiment data, the training error margin that get is 9.99635×10-5, test the error margin as 1.7065×10-4.7. Use the Visual C++6.0 languages carried out the function of above- mentioned each link, developing to identify to count the system software package, equipping to match each other to match with the hardware that research to manufacture, 0-500 of germs quantity that a picture can collect, error margin±2%.Be no bigger than the 30 s to analytical time of one picture.8. The contrast experiment. Apply to statistics to learn the principle, carry on this system and traditional methods comparison, in germ liquid, milk, juice and the contrast of the beef experiment, the result difference that two kinds of method measure is not obvious.The rapid detection on the total bacteria is not reported at present. This paper has tested at first. Though there are also many disfigurements on hardware design, the result is satisfaction for us. Because of the rapid detection on the food total bacteria is the most important work in quality control stations in many provinces, towns and counties. This system can develop very well in the market. In the future, we must improve on the aspect of speed, characteristic pick-up, hardware design, enhance exactness, and sale promotion this system to the market. |
PDF Full Text Request