Design And Implementation Of Cross-Platform Library For Microscopic Image Analysis Of Algal Blooms

In recent years, red tide occurs more and more frequently and brings great harmto marine ecology, marine economy and human health. It causes extensive attentionboth at home and abroad. In order to forecast and prevent the occurrence of red tide,our country has studied the occurrence mechanism of red tide in multiple directionssuch as fundamental researches and hi-tech development, and began to set up anoperation based detection system of red tide. Effective and efficient identification foralgal blooms is an important part of such a system.Using microscope to recognize the algal blooms artificially is a highlyprofessional and time-consuming work. Aiming at solving this problem, this paperdesigns a library for microscopic image analysis of algal blooms according to thetechnology of computer image processing, and introduces the implemention on PC,Android, and Web platform. It can be used in the form of cross-platform applicationby red tide algae researcher. The library for microscopic image analysis of algalblooms includes two modules, image processing and patterns recognition. It providesa function library for microscopic image of algal blooms and a set of tools for patternsrecognition. The interface implementations are all based on open source library(OpenCV) and libsvm.PC platform implementation for microscopic image analysis library of algalblooms is designed by Code::Blocks and wxWidgets. It provides functionalities suchas image processing, patterns training and algal blooms prediction. Android platformimplementation is designed by Google Android SDK and NDK. Researchers canmake use of image processing and algal blooms prediction at any time via usingAndroid mobile devices. Web platform implementation is designed by LAMP architecture and ThinkPHP framework. It has been shown on "Diagnostic Platform ofHarmful Algal Blooms" website which belongs to national "863" project named"Study on Diagnostic System of Harmful Algal Blooms". Researchers can login"Diagnostic Platform of Harmful Algal Blooms" website and access the microscopicimage recognition page in online diagnosis module, then make use of the onlinediagnostic functionality of microscopic image recognition.At last, this paper selects4000microscopic images of algal blooms to do anexperiment. The results show that: The microscopic image analysis library of algalblooms has a high rate prediction, and the cross-platform software is complete andstable. It’s a good tool of automatically predicting for algal blooms and has profoundsignificance for the prevention of red tide.