New scheme for algae assisted identification and counting

In 2009, Xunquan launched the world's first "image-based plankton detection and intelligent identification system" to cater to the country's emphasis on environmental monitoring, providing an effective means for algae monitoring and research in environmental monitoring agencies and research institutes.

In the algae identification process, it is generally based on the specific morphology of the algae. Because the morphological characteristics of the algae are complex, the morphology exhibited by different angles is different at different times, so the identification of algae brings a lot of problems. . In the original algae system, the morphological search-based identification system pioneered by Xunshu, combined with the algae expert database for search and identification, can effectively narrow the range of algae and reduce the workload.

In the new generation of algae counting system launched in 14 years, two new "Xunxun" core technologies have been added: high-precision intelligent search for biological similarity, chaotic intelligent classification and counting, rapid algae-assisted identification and automatic segmentation of different algae. Biosimilarity High-precision intelligent search is the core technology of the new generation of algae intelligent identification. Through the effective combination of “morphological similarity” and “biological similarity”, the bio-characteristics of algae species are accurately extracted and integrated, and algae are greatly improved. Search accuracy makes rapid algae identification possible.

principle:

1) Color feature extraction: The color characteristics of algae cells are extracted according to the color histogram.

2) Texture feature extraction: Based on the rotation invariance feature of Gabor filter, the texture characteristics of algae cells are extracted.

3) Intelligent search: The two features are merged into feature vectors, and the classifiers of the support vector machine are used for training to realize the classification and search of the algae cell images.

Example:

Automatic classification and counting of mixed algae: Chaos intelligent classification and counting is a major technological breakthrough in the research of algae automatic classification and counting. It has initially realized the automatic classification and counting of many types of algae cells with large differences in morphology and color.

principle:

1) Chaotic genetic algorithm: Using the randomness, ergodicity and initial value sensitivity of chaotic motion, the chaotic state is introduced into the optimization variable, and the traversal range of the chaotic motion is extended to the range of the optimization variable, thus achieving Image segmentation of all algae in the water.

2) Fuzzy C-means clustering algorithm: Determine the degree to which each algae sample data belongs to a certain cluster, and classify the algae with similar degrees of membership into one cluster.

Example: