Pixel-Based Leaf Segmentation from Natural Vineyard Images Using Color Model and Threshold Techniques

摘要

The presence in natural vineyard images of savage foliage, weed, multiple leaves with overlapping, occlusion, and obstruction by objects due to the shadows, dust, insects and other adverse climatic conditions that occur in natural environment at the moment of image capturing, turns leaf segmentation a challenging task. In this paper, we propose a segmentation algorithm based on region growing using color model and threshold techniques for classification of the pixels belonging to vine leaves from vineyard color images captured in real field environment. To assess the accuracy of the proposed vine leaf segmentation algorithm, a supervised evaluation method was employed, in which a segmented image is compared against a manually-segmented one. Concerning boundary-based measures of quality, an average accuracy of 94.8% over a 140 image dataset was achieved. It proves that the proposed method gives suitable results for an ongoing research work for automatic identification and characterization of different endogenous grape varieties of the Portuguese Douro Demarcated Region.