Quick Quantitation System For Bacteria Number In Fresh Milk

With the continuous improvement of living standards, people on the dairy increasinglyare stringent quality requirements, and the bacteria number of dairy product is a key aspectto measure the quality of dairy product. As a result, bacteria number quantitation methodsbecome a significant research field. Although many methods, such as bacteria plate count,?ow cytometry and impedance method have been broadly used in the dairy industry toquantitate bacteria number around the world, none of them is a quick, low cost and easy one.As a result of these various reasons, some testing agencies switch to manual quantitationmethods. They first get the sample of liquid, then use microscope to observe and get images,and count number of bacteria. However, this method is not suit to large-scale industrialproduction. In this case, we use image processing method to quantitate bacteria number infresh milk in food industry without dying the sample liquid. This system is mainly used totesting bacteria number level of fresh milk. This test will occur when farmers deliver rawmilk to dairy plant. The main purpose of testing is helping milk factory grading raw milk. Inthis project, Because the speed is important in industry, the enlargement ratio of the imageof the sample will be small.In order to raising the representativeness of samples to reduce the number of sampling,we must expand the field of view of microscope. But it also reduce the magnification of themicroscope, which will reduce features of bacteria of image, so, obviously we need to findbalance between the two. At the same time a smaller enlargement ratio made bacteria ofimage contained only a few pixels. The article first gave a briefing on some of the existingmeasurement methods, and then described algorithm used to recognise bacteria, including aspecial threshold selection algorithm, a method to remove the points of misjudgment. Thenstatistical inference and database module are designed. XP programming is also introducedwhen talking about software engineering methods using in the project. Finally, in themulti-core CPU, I discussed how to use parallel programming to optimize the software.The recognition rate of bacteria in images are good, and the optimization for multi-coreprocessors has also been made in the expected results.