Application of Computer Vision for Lettuce Seeds Germination Detection

This paper presents a unique application of computer vision methods designed to detect germination of lettuce seeds under controlled environment. Two steps involved in this application are image acquisition and image analysis which uses image skeletonization and other image processing algorithms. Seed images are converted to binary images followed by extraction of seed skeletons, which are used to calculate the changes in seed length. Other parameter such as area is also calculated. Seed area and length changes towards radicle emergence over time are used to identify the onset of germination. In addition, when integrated in germination test, results of our approach show that efficiency on seed quality assessment may be improved.