Portable device for contactless, non-destructive and in situ outdoor individual leaf area measurement

Abstract

Plant phenotyping is a research area concerned with the quantitative measurement of a plant’s structural and functional properties. In the case of measuring a leaf’s surface area, it is more often than not laborious as well as stressful to the plant. In this paper, we present the use of the RGB-D sensor, Kinect v2 as part of a portable device for non-destructive measurements of individual leaf areas (cm2/leaf) outdoors in daylight. The Kinect v2 was utilized to capture a single viewpoint 2.5D frame of plant foliage. An unsupervised clustering method, HDBSCAN was used to segment out individual leaves from the captured 2.5D frame of the subject plant. Performance of the leaf segmentation was measured by evaluating the 10 nearest (max) clusters from the sensor for each frame into 3 different categories, individual leaves (non-occluded, occluded), under-segmented and over-segmented. Probability of segmenting individual leaves differs from plant to plant, ranging from a low of 0.7178 to a high of 0.8975. The surface area of all individual non-occluded leaves obtained via the segmentation method was calculated and compared to its ground truth. The calculated individual leaf surface areas R2 was recorded to range from 0.792 to 0.911 with respect to its best fit regression line while the RMSE range from 4.9482 to 14.4941 cm2. The proposed system and method was shown to be capable of segmenting individual leaves from dense foliage and measuring its surface area.