The Image Segmentation And Feature Extraction Of Plankton Diatoms

The marine planktonic algae species are important primary producers in the ocean, which are the food of marine animals directly or indirectly. What's more, the algae plankton species can concentrate pollutants. Researching the algae plankton species has been drawn more and more attention in ocean fishery, marine environment protection and so on. Traditionally, in the study of algae plankton species, different technologies have been proposed, such as microscopic inspecting technology, shape analysis as well as statistics methods,which are the most fundamental and common. However, those methods mainly depending manual operation which are easy to make mistakes, have the disadvantages of labor-intensive operation with inefficiency. Meantime, it is not easy to save and progress the data. With the development of computer technology and digital image processing technology, making use of image processing technology to realize the automatic recognition and analysis for the algae is the trend in the future. Nevertheless, due to the complexity and diversity of algal cell as well as the extreme complication of texture inside, the common image segmentation in the alga image processing is very difficult to segment efficiently. The main objective of this study was to research the methods of image segmentation and the feature extraction of diatom cell. In addition, with regard to the feature extraction of diatom cell, a whole solution has been proposed to save those problems, which establishes the good foundation of subsequent algae image classification and recognition.In the first instance, the magisterial thesis introduces the research and development of algae image analysis and recognition together with image segmentation at home and abroad. All those methods have been analyzed and compared. The advantages and disadvantages of those methods have been concluded as well as the problems existed. Because of the complex natural quality and structure characteristics feature of diatom cell, the traditional image segmentation methods have the drawback barely conquered. After a great deal of experiments, the method based on C-V segmentation to segment the algal cell image has been proposed. The method has obtained a very good effect. The result future proves the feasibility and efficiency of this method. Based on the image segmentation, the characteristics extraction of algal cell has been done in figure, color and texture. In the figure, several characteristics have been extracted, such as squareness, the value of arc and curvature. Meantime, the method of color histogram has been employed to extract the color characteristic. At the same time, the similarity also has been calculated. In the feature extraction, Gray-level Co-occurrence Matrix has been adopted to extract the texture features of cell image. All results of those experiments have proved the feasibility and efficiency of this method.The application software base on VC, which owns the function of image preprocessing, image segmentation of algal cell and the feature extraction in figure, color and texture, has been designed and developed. Because of it, the high precision, rapidity, automatical microscope image segmentation of algal cell and feature extraction have come true.