Maize Stem Contour Extraction and Diameter Measurement Based on Adaptive Threshold Segmentation in Field Conditions

Solving the problem of stem contour extraction maize is difficult under open field conditions, and diameter cannot be measured quickly nondestructively. In this paper, at small large bell stages was object study. An adaptive threshold segmentation algorithm based on color space model proposed to obtain in field. Firstly, 2D images were captured with an RGB-D camera. Then, processed by hue saturation value (HSV) space. Next, extracted maximum between-class variance (Otsu). Finally, reference method used maize. Scatter plots Dice coefficients compare effects HSV + fixed algorithm, Otsu K-means algorithm. The results showed that optimal choice for extracting contour. mean absolute error, percentage error (MAPE), root square (RMSE) stage 4.30 mm, 10.76%, 5.29 respectively. MAPE, RMSE 4.78 12.82%, 5.48 MAPE within 10–20%. could meet requirements measurement provide a acquisition phenotypic parameters meantime, conditions provides technical data support precision farming plant breeding.