Double Spectrum Based Image Analysis Of Harmful Algae

The red tide phenomenon has long caused the attention of the sea-sided nations and the related fields of science. Plans on red tide researches have been made accordingly. In all those research works, studies of the morphological and spectral characters of the harmful algae are the most fundamental and important portion. Currently, the most common method for the analyzing of the harmful algae is to manually sorting and counting algae under microscope. Evidently, this method is very primitive and inefficient due to the affection of the human subjective factors. Thus, the development of double-spectrum based harmful algae analysis instrument for lab use and the setting of morphologically based image database become a meaningful work, which will set a good foundation for the research of automatic harmful algae analysis instrument for fieldwork. There are nationally or internationally no reports about the mature technique or instruments concerned. Combined the characters of fluorescence, morphology and texture, double-spectrum based technique can analyze the features of harmful algae more effectively and exactly. Besides, it will be a great help for the discovery of new type of algae by applying the morphological image database into the harmful algae recognizing and sorting. According to special characters of the red tide, the subject developed completed system for the analysis of harmful algae image, including fluorescence and gray micrograph sampling system and image processing software. On the basis of software and hardware, we have realized the analysis and recognition of six kinds of algae, and set up an image database of harmful algae, which runs beyond the primary requirement of the schedule. The design of micrograph sampling system is composed by three parts: the choosing and switching structure design of the fluorescence inspiring light, the reconstruction of the microscope for the use of fluorescence and black-white imaging capture and micrograph screening by CCD or digital camera and its transmission. The software includes six parts: image capture, image preprocessing, character acquiring and analyzing, database setting and sorting method research. Firstly, we design control system according to the character of the algae to acquire the micro photo, then we erase the noise and enhance the quality of the image before the automatic threshold segmentation and mathematic morphologic procession. After that, we acquire 15 characters of the harmful algae, including characters of texture, shape, and black-white value. In addition, we proposed two-step analysis method to save time and enhance efficiency of the analysis. Finally, we set up an image database to record the methods that used in the procession of harmful algae image and the specific features obtained from the analysis. We have taken experiment on six typical harmful algae using our system. Those algae are Gymnodinium Stein, Saline Algae, Noctiluca scientillans, Heterosigma akashiwo, Coscinodiscus gigas Ehrenberg and PR Algae. The result shows that it is possible for the automatic sorting and counting of certain kinds of algae.