Flatfish Measurement Performance Improvement Based on Multi-sensor Data Fusion

In this study, a multi-sensor data fusion system using a load cell and vision sensor was considered in the development of a flatfish classifier for systematic fish management in aquaculture. In the single-sensor measurement method, each sensor has disadvantages. A load cell shows high performance in the measurement of adult fish, but the measurement of fry is affected significantly due to water weight (water weight disturbance). A vision sensor shows high performance in the measurement of fry, but the movement of fish (movement disturbance) affects the accurate measurement of adult fish. Therefore, in this study, these disturbances were compensated for using a datafusion algorithm, of which the performance was evaluated by a comparison between single sensor measurements and multi-sensor data fusion results.

Automated summary

This study considered a multi-sensor data fusion system using a load cell and vision sensor for a flatfish classifier in aquaculture. The load cell performs well in measuring adult fish, while the vision sensor is better for measuring fry, and a data fusion algorithm was used to compensate for their respective disadvantages in fish measurement. The performance of the system was evaluated by comparing single sensor measurements with multi-sensor data fusion results.