In this study, we explored the driving factors behind plankton community structure. Due to the rapid development of cities, the occupation and development of wetland resources have increased lately, making the urban wetland ecosystems unstable and degrading the ecological functions gradually. Understanding the driving factors behind plankton community structure has certain theoretical and guiding significance for the protection, sustainable development, and ecological restoration of aquatic biodiversity in urban wetland ecosystems. We set up 12 sampling points in the Hulanhe Wetland, with the continuous monitoring of plankton from April to August and October 2021. The eco-environmental factors, plankton community structure, biodiversity index, resource use efficiency (RUE), and Bray–Curtis community turnover value were analyzed. A total of 209 species of 91 genera, 42 families, 11 classes, 22 orders of phytoplankton, and 90 species of four classes of zooplankton were identified. The community structure was mainly composed of Bacillariophyta, Chlorophyta, Cyanophyta, Protozoa, and Rotifera. To explore the correlation between phytoplankton and zooplankton, a correlation study was performed. We found a stable feeding preference between phytoplankton and zooplankton. The key influencing factors were identified by ordinary least squares regression, and the main driving factors of plankton community structure were discussed. The results showed that the stability of the Increased biodiversity and resource utilization efficiency have led to more stable plankton communities. This stability pattern is also strongly affected by water temperature, pH and total nitrogen in the external environment. This study will be helpful in the restoration of damaged wetlands, which would be beneficial for the protection of urban wetland ecosystems.

Aiming at the problem of low traffic sign recognition rate and slow speed, a traffic sign recognition algorithm combining CNN and Extreme Learning Machine is proposed. First, the ResNet50 network is used to extract image features, and then the Region Proposal Network (RPN) is used to generate proposals from the extracted image feature maps. Finally, the extreme learning machine is used to classify the generated proposals, and the fully connected layer is used for regression prediction. The experiment shows that compared with the Faster R-CNN model, the CNN+ELM improves the recognition accuracy on the TT-100K dataset 7.7% and reduces the training time per epoch by 32 seconds.

In order to solve the problem of low matching precision and slow matching speed of image matching algorithm based on classical SIFT, a new image matching algorithm based on PCA, SIFT and improved RANSAC is proposed. Firstly, the SIFT feature was extracted from images; Secondly, principal component analysis is used to reduce the dimension of SIFT feature descriptor, from 128 to 20; then, the EUCLIDEAN distance is used for feature matching; finally, an improved RANSAC algorithm is proposed to eliminate the mismatched feature points. Experimental results show that the proposed algorithm improves the accuracy and speed of image matching.

No abstract