Image Processing Research Of Phytoplankton Based On Multi-wavelet Theory

As the most important primary producer in the marine ecosystems, phytoplankton influences the marine environments and the living marine resources enormously. How to classify and recognize the phytoplankton cell is the focus of our research. Based on multi-wavelet theory, the de-noising and feature extraction of the phytoplankton cell micrograph are studied in this paper mainly. The paper mainly included following aspects:1. Based on wavelet history and theory, this paper has introduced the development of multi-wavelet and theory, at the same time, introduced the four characteristics of multi-wavelet. According the desire performance of the multi-wavelet characteristics, we introduce a approach of structuring orthogonal symmetric multi-wavelet.2. To improve the quality of image and facilitate image further processing, recognition and so on, it is necessary to de-noise image. The paper has improved traditional Donoho threshold algorithm, multilevel threshold and shrinkage-function. In order to overcome the Pseudo-Gibbs phenomenon, we used the method of translation invariance. The experimental results show that the de-noising method adopting the new threshold function gives better MSE performance and SNR gains than hard and soft thresholding methods. However, the hard-threshold is best in preserving edges but worst in de-noising, and soft -threshold is best in reducing noise but worst in preserving edges, this method incorporates the hard and soft thresholding to achieve a compromise between the two methods.3. Feature extraction is the key factors in classification. According to multi-wavelet multiresolution analysis characteristic and texture characteristics of Coscinodiscus, we have proposed a method of extracting feature based multi-wavelet and PCA, and recognize the cells with 3- neighborhood classification. The experimental results show that the new method not only can improve the recognition rate, but also can greatly reduce the computing time and increase the computing speed.Through the research of phytoplankton image processing, we have improved the image quality of phytoplankton cells, effectively analyzed and extracted of the characteristics of the phytoplankton cells, reduced the computing time and increased the recognition rate for the classification, which lays a foundation for the classification and the recognition of the phytoplankton.