A feature extraction approach based on Hough transform to estimate growth responses of leaf lettuce

摘要

Machine vision was a conventional tool which has been commonly used to obtain morphological features of plant image through image processing method. This study proposed a machine vision-based image processing approach to identify the number of leavesand contours of leaf lettuce. First, the entire image was pre-processed through color conversion, binarization, median filtering, denoising, and thinning operations to obtain the contour of lettuce and classify the main vein of leaves. Then, the featureextraction method based on Hough transform was utilized to identify the number of leaves. In addition, central area of lettuce detection and connected-component labeling operation were also employed to count the number of leaves. The embedded micro-computer was utilized to acquire the image in real-time and estimate the number of leaves and the area of leaves with pixel in the contour of leaves, whose identification results were compared with the experimental data. The calibration models with linear regression were generated to relate the number of leaves and dry mass of lettuce to data measured by a grower. The results illustrated that the coefficients of determination of proposed methods were greater than 0.91, indicating the effectiveness of the method.