Digital camera based measurement of crop cover for wheat yield prediction

Abstract

Quantification of vegetation cover can facilitate an improved understanding of ecosystem function and crop yield. Current methods of cover measurement can be divided into field surveys and remote sensing, with their associated advantages and disadvantages. Digital camera images provide high temporal and spatial resolution at low cost than satellite images. A method to accurately and conveniently estimate crop cover from digital camera images is presented. The digital camera images were converted from red-green-blue to hue-saturation-intensity color space. Hue segmentation technique enhanced the characteristics of plant tissues, and identified green tissues easily. Vegetation and non-vegetation binary pseudo-color images were produced and crop cover was calculated for each plot. By comparing leaf area index and wheat yield with time series cover data from the 2005 growing season, two results were obtained: (1) there were high correlations between cover and leaf area index at four stages (r2 = 0.88, 0.96, 0.97, and 0.92); (2) crop cover at 80 - 90 days after sowing can be used to predict the wheat yield with good results. Our results suggest this method is an accurate and efficient means of measuring vegetation cover in cropland.