Developing image processing techniques for estimating eggplant trichome density
2019
Porca, A.A. | Madrid, V.R.M. | Taylo, L.D. | Cainday, J.T. | Hautea, D.M.
Eggplant leaf trichomes have been documented to be an insect resistance factor especially for sucking insects. Trichomes serves as a mechanical barrier that impede feeding and oviposition of leafhopper. Characterization of trichomes of eggplant germplasm is labor-intensive and prone to bias. This study explored the implementation of a digital solution using computer vision to automatically count the density of leaf trichomes and the trichome images used were taken at the Institute of Plant Breeding on February 2019. The automation was divided into four parts: (1)image acquisition, (2)image enhancement, (3)image binarization, and (4)trichome cluster detection and counting via its central disk. First, the leaf images were acquired using a stereo zoom trinocular microscope with 20x magnification. Second, the images were enhanced by using a sharpening and histogram equalization techniques. Third, the enhanced images were then subjected to grayscale morphology opening operation to bring out the central disks of the trichomes. Then the images were binarized using intensity slicing and adaptive thresholding to isolate the central disks of trichomes. Finally, each cluster of trichome was detected and counted if they were found inside the leaf structure in the image. The computer vision approach developed was able to provide estimate with 85% degree of precision. Although only stellate trichomes were analyzed in this study, the technique used can be easily modified to detect other types of trichomes in other crops as well.
显示更多 [+] 显示较少 [-]