Herein, an accurate and efficient algorithm for digital-display instrument positioning and recognition is proposed. The isolated forest algorithm and Otsu watershed threshold algorithm were used to distinguish digital...
详细信息
Herein, an accurate and efficient algorithm for digital-display instrument positioning and recognition is proposed. The isolated forest algorithm and Otsu watershed threshold algorithm were used to distinguish digital-display instruments from nondigital-display instrument areas and separate the foreground from the background, respectively. The histogram of oriented gradient??? support vector machine classification algorithm was used to distinguish instrument and non-instrument regions, which considerably improved the accuracy of digital-display instrument region positioning, avoided the interference of non-digital tube character regions, and reduced the search time of the digital tube region. A convolutional neural network was used for character recognition. Global characteristics of the character region were fully utilized, and partial digital character issues and scenarios where the decimal point is not obvious were mitigated. The proposed method can adapt to angle deviation, partial character missing, and image noise and exhibits excellent robustness and adaptability to the location and recognition of the digital tube.
In order to improve the efficiency of urban drinking water safety monitoring and early warning management, a pollution risk early warning model of urban drinking water supply chain is proposed. Firstly, the current si...
详细信息
In order to improve the efficiency of urban drinking water safety monitoring and early warning management, a pollution risk early warning model of urban drinking water supply chain is proposed. Firstly, the current situation of urban drinking water supply is analyzed and the causes of pollution are analyzed. Then, the AR model (autoregressive model) is used to predict the time series of multiple water quality indicators by constantly introducing new monitoring data mode for the residual vector group, the outlier scores of each vector group are obtained by using the isolated forest algorithm to judge whether the water quality is abnormal or not, and the fuzzy comprehensive evaluation method is used to evaluate the level of the abnormal situation and carry out the corresponding level early warning. The experimental results show that the AUC can reach 0.919 when using the prediction residual vector group of turbidity and conductivity to detect the numerical changes of water quality parameters in drinking water supply chain, accurately predict the abnormal data, make early warning, and provide guarantee for the survival of urban residents and urban development.
In the service dimension, the construction of fitness science data supervision service mode is discussed. Based on the stakeholder theory, through the statistical analysis of the stakeholders of fitness science data s...
详细信息
In the service dimension, the construction of fitness science data supervision service mode is discussed. Based on the stakeholder theory, through the statistical analysis of the stakeholders of fitness science data supervision, three core stakeholders of the government, users and data service personnel are identified. Based on these three dimensions, we find out the core concepts of government policy model, user demand model and service model. At the same time, each dimension is deeply analyzed. Through the relationship analysis between these three dimensions, the user-oriented collaborative supervision service model of fitness scientific data is expected to guide the specific service practice of fitness scientific data supervision through the establishment of this model. In addition, an unsupervised learning method in machine learning, the isolation forestalgorithm, is introduced to detect abnormal data;at the same time, using real fitness data sets, through comparative experiments with local anomaly factor algorithms, it is verified that the isolation forestalgorithm has a good effect of anomaly detection;this article also uses redis cache to optimize the performance of the fitness data monitoring system, which solves the access pressure of the main database in a multi-user high-concurrency environment;Finally, the usability and stability of the system are verified by functional tests and stress tests.
暂无评论