The Outside Contour Extraction And Shape Analysis In Marine Phytoplankton Automatic Identification System

Marine phytoplankton is not only important marine biological resources, but also perpetrator of red tide and other natural disasters. The resources utilization and the monitoring of red tide on marine phytoplankton can not be separated from Marine Phytoplankton classification and identification. Processing of classification and quantitative analysis usually manipulated by manual under microscope by far which is time-consuming and laborious, however, it is one of the most effective way currently used. It is expected to form a set of phytoplankton automatic identification system depends on microscope and computer scanning system if we can establish a characteristics database of various phytoplankton and develop a set of data-processing software. It will have great signification to the identification of Marine phytoplankton.The main study of this thesis is the outside contour extraction of marine phytoplankton images ,and their recognition and classification based on some shape characteristics. At First ,the thesis uses the maximum area to locate the object, refering the characteristics of the Marine phytoplankton image . And based on that, PCNN is used for filling the internal region of the object and getting the outside contour of the object. Secondly, the thesis introduces some common characteristics of the shape and hierarchical classification: the thesis describes how to use the Hausdorff Distance between phytoplankton to determine the number of rotation axes and apply it to distinction the major categories of phytoplankton fistly, secondly it also introduces other category classification between species.in use of its' own shape characteristics ; Besides the thesis introduces how to use SVM to do classification and identification of each categories, which also introduces kernel function selection, parameter selection, as well as K-fold cross-validation..At last, the thesis analyses the effects of identification in our system, summarizes the innovation work and put forward some future work on the systemThe thesis take the extraction method of location firstly and then extraction,when extracting the outside contour;In the extraction of shape features, the number of rotation symmetry axes is used to classify Marine phytoplankton images.