TY - CHAP
T1 - Computer Vision for Recognition of Fruit Maturity in Amazonian Palms Using an UAV
AU - Marín, Willintong
AU - Colorado, J.
AU - Bernal, Iván Mondragón
N1 - Publisher Copyright:
© 2020, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - This paper presents the integration of well-known computer vision methods to identify the stage of maturity of the Asai, Seje and Moriche fruits in Amazonian palm based on aerial images acquired with an Unmanned Aerial Vehicle (UAV). Despite the aim of this research is to use both multispectral (NIR) and thermal cameras to acquire imagery at different wavelengths, this paper is limited to a maturity classification by extracting features from visible spectrum imagery (VIS). We have implemented an algorithm that combines the VIS image histogram with a mask filter and the corresponding thresholding to process the acquired images. A classifier is being used to recognize the maturity stage of Moriche palm fruits. Classification results have shown an overall accuracy of 66.5% and a performance of 58.33%. These preliminary results confirm that we need to include NIR information for enabling the extraction of more relevant features related to the fruit maturity stage. An approach based on NIR vegetative indices will be implemented in upcoming work.
AB - This paper presents the integration of well-known computer vision methods to identify the stage of maturity of the Asai, Seje and Moriche fruits in Amazonian palm based on aerial images acquired with an Unmanned Aerial Vehicle (UAV). Despite the aim of this research is to use both multispectral (NIR) and thermal cameras to acquire imagery at different wavelengths, this paper is limited to a maturity classification by extracting features from visible spectrum imagery (VIS). We have implemented an algorithm that combines the VIS image histogram with a mask filter and the corresponding thresholding to process the acquired images. A classifier is being used to recognize the maturity stage of Moriche palm fruits. Classification results have shown an overall accuracy of 66.5% and a performance of 58.33%. These preliminary results confirm that we need to include NIR information for enabling the extraction of more relevant features related to the fruit maturity stage. An approach based on NIR vegetative indices will be implemented in upcoming work.
UR - https://www.scopus.com/pages/publications/85079288020
U2 - 10.1007/978-3-030-40309-6_4
DO - 10.1007/978-3-030-40309-6_4
M3 - Chapter
AN - SCOPUS:85079288020
T3 - Lecture Notes in Networks and Systems
SP - 31
EP - 39
BT - Lecture Notes in Networks and Systems
PB - Springer
ER -