Smart Morphological Change Sensing System using an Automatic Non-destructive Method for Soybean Flower Bud Differentiation

Most of the research on the flower bud differentiation is based on manual measurement. In this paper, a smart morphological change monitoring system using an automatic non-destructive method for soybean flower bud differentiation is presented. A non-destructive measurement system using multiple cameras to measure the morphological characteristics of soybean flower buds was developed and constructed. An automatic threshold segmentation algorithm (OTSU) was used to segment a flower bud image to obtain a binary image, and the morphological parameters of the flower bud characteristics were obtained. The results demonstrated that the optimal reference plate size in the algorithm was 220×230 pixels, and which can adapt to 60h time matching and the NCC was higher. Compared with the manually measured results, the coefficients of the flower bud area and perimeter obtained by our proposed monitoring system were 0.9611 and 0.8983, respectively. In addition, the relative errors of these values were less than 5.7% and 4.4%, respectively, while the relative accuracy of the measurement system could reach 0.01 mm. It was proved that the real-time monitoring method for flower bud differentiation greatly eases the measuring workload and overcomes the limitations of manual measurements.