Research On Image Analysis Of The Harmful Algae

As is known,red tide is a serious global marine disaster causing billions of dollars' economic loss to the world and at least over 1 billion yuan loss to China. Many methods had been researched for prediction of the red-tide disaster. The inchoate identification and measurement of the harmful algae is the main way to predict the red-tide. At present, the identification and measurement mainly observed and analyzed in the lab artificially. But the artificial observation and analysis can not satisfy the prediction of the red-tide because of the huge intention of the work. So it is an impending task that we develop an automatic analysis equipment for the harmful algae.This paper studied the pre-process, segmentation, feature extraction and classification of the harmful algae's image based on the characteristic of the harmful algae's micro-image. Firstly, we analyzed the origin and characteristic of image noise and designed the method to remove the noise. The theoretic analysis and the results of the experiments proved that 2D entropic threshold algorithm could segment the harmful algae's image effectively, compared with the other auto-segmentation methods. The mathematical morphology was applied to solve the problem of the overlapping algae. We solved the problems of feature extraction and classification by the method of object identification using the boundary and the texture.According to the principle of the microscope measurement, we designed an experimental way for the application. Finally, the experiments of the normal granule and Noctiluca Scintillans proved that we can measure the deepness and identify the harmful algae by this way.