Research On Red Tide Algal Cell's Recognition

Red tide algae is a main creature which induces the red tide frequently. Red tide not only do some harm badly to the circumstance around the ocean and fishery, but also affect people's health and life disadvantageously. Therefore, red tide algal cell's correct recognition is very important for availably preventing and identifying the ocean red tide's happening.Red tide algal cell's recognition system is made of four parts which include image's pretreatment, image's segmentation, extraction of image's characters and image's classification. Digital image processing is used in the dissertation, and the image recognition system's four parts is brought in. The research on the red tide algal cell images'recognition is summarized below:1. Image's pretreatment: Image's gray level is amending and it's noise is wiped off. Mathematical morphology is taken to wiped off the noise of red tide algal cell's binary image.2. Image's segmentation: A new method which combine multi-scale mathematical morphology with tagging image's connected components and edge detection operator is put forward to extract the boundary of red tide algal cell. The result of the experiments shows that the method is simple to carry out and the boundary extracted has smaller distortion.3. Extracting image's characters: Algal cell's roundness, rectangle-rate, elongation-rate and wavelet moment is extracted. Image's unitary shape character can be described by the characters above. It's local shape details can be described as well.4. Image's classification: Modular neural networks is used for red tide algal cell's pattern recognition, and the recognition ratio is above 95%. The results of the experiments have shown that the method can recognize the red tide algal cell more accurately.