Abstract: The shadow and halo of leaf edge have great influence on measurement accuracy of leaf area. In this paper, the causes of the shadow and halo were analyzed and a correction method was proposed. Firstly, the image color space was transformed from RGB to Lab, and the rough leaf region was abstracted by using the adaptive region growing method. Secondly, the result of region growing was processed by morphological dilation and erosion and the exact leaf area, background area and unknown area were segmented. Finally, the pixels of unknown area were classified to the leaf area or the background area with the improved MRF. Experiments show that this method can greatly improve the accuracy of leaf area measurement.

Key words: image processing, leaf area measurement, Lab color space, region growing, MRF