Fractal and Recognition Method of Underwater Zooplankton Image
Due to its large land area and rich species, the ocean has an irreplaceable role in the earth’s ecosystem. Among them, marine zooplankton is a complex and very large biological population, which has important research significance in many important fields such as new gene discovery, marine environmental protection and climate change. How to effectively observe the changes in the number and distribution of marine zooplankton which plays an extremely important role in marine ecosystems, However, due to the variety of marine zooplankton species and various forms, traditional collection by artificial biological trawling, and then relying on manual analysis and statistics. The method is very inefficient and does not have a good recognition effect, and it is difficult to meet the needs of today’s zooplankton research. Therefore, this paper combines the idea of image processing to study how to identify underwater zooplankton by underwater image of zooplankton. Firstly, the image blur enhancement algorithm using maximum fuzzy entropy is used to enhance the collected zooplankton image, and the image quality is enhanced while retaining the original image information. Secondly, based on the fractal theory, fractal features such as fractal dimension, void and multi-fractal are extracted, and the eigenvectors are combined with gray features. Finally, using the composed feature vector as input, the convolutional neural network is used for classification and recognition, which effectively realizes the recognition of underwater zooplankton. The simulation results show that the image fuzzy enhancement algorithm based on the maximum fuzzy entropy designed in this paper can improve the quality of acquired images and facilitate the extraction of fractal features. Further, using the convolutional neural network to identify the feature vectors composed of fractal features and gray features; it can have higher accuracy and can realize the recognition of underwater zooplankton.