Growth Analysis And Select Technology Of Yeast Based On Digital Image Processing

In recent years, the research on the yeast has become a hot spot in many fields, such as liquor‐making industry, medicine manufacture, and agricultural fermentation. Usually the survival rate, budding rate and strong rate reflect the fermentation capacity of yeast. It is the morphological analysis and counting statistics that is the most commonly way to identify the quality of yeast. Traditionally these tasks are done manually, so it makes the results less accurate and high cost.According to the requirements of yeast research on the fermentation capacity, this paper established technology research to analyze the yeast based on digital image processing technology. The following aspects item has been mainly researched:1) Based on the analysis for characteristics of yeast growth and physiological, we select three standards identify the fermentation capacity of yeast. These three elements are survival rate, budding rate and robustness rate. These jobs laid the foundation for subsequent research.2) Established a method to automatic detection and analysis the rate of yeast. Accorded to thegrowth pattern and the morphology of yeast, a ellipse fitting method which is based on concavepoint detection is proposed. This method is used to analysis the rate of yeast by the outline of thegrowth process. This algorithm could detect the bud-rate of the yeast exactly even if the cells wereaccumulated. This method is not only fast and more accurate when compare with traditionalmethods.3) Established a method to detection the survival rate of yeast. Based on color similarity, proposed discriminates analysis of yeast whose dead cells could be dyed by biological agents. Experimental results show that the new detector is more suitable for yeast‐survival‐rate.4) This paper is proposed a method of screening good yeast. Accorded to the internalorganizational structure differences between them, this paper extracted the texture features of strongyeast and senescent yeast by GLCM. We got the data like energy, correlation, entropy and contrastform a lot of experiments, and contrasted them. At last we formatted a yeast robustness analysis toidentify the good bacteria and bad bacteria. This paper verified the feasibility and effectiveness ofthese methods, by comparing a large number of experiments and analysis.This paper verified thefeasibility and effectiveness of these methods, by comparing a large number of experiments andanalysis.