Identification of Replantation Areas in Sugar Cane Rows using Aerial Imagery
We have developed an algorithm that highlights the regions that need to be replanted in a sugar cane crop using aerial imagery in the near-infrared band. Under this band, chlorophyll and vegetation are contrasted better with the soil. With this information, we can better determine the distance between sugarcane crops in a row.
The algorithm segments each sugar cane, calculate the distance between sugarcanes in a row and finally highlight the region of interest for replantation in order to improve the productivity in each growing cycle.
Sugarcane is an important economic resource for many tropical countries. Studies say gaps between crop regions correlate with planting quality. [Luna and Lobo, Remote Sens. (2016)]
Best way to maximize yield is to minimize the occurrence of gaps between crop regions
Currently, farmers use sensors mounted on a truck or use manual sampling, these methods are labor and time intensive.
The goal of this project is to exploit the capabilities of image processing without the use of any learning method (eg. SVM, NN, regression, etc.). Our algorithm was implemented in C using VisionX library.
The algorithm includes methods such as filtering, threshold, moments and transforms. The main contribution was the development of a customized Hough Line Transform that identifies the direction and location of the crop rows in the field and its inverse. Additionally, a custom filter was applied in our custom Hough Space to improve the analysis.
This work was developed in collaboration with Jonathan Wu (firstname.lastname@example.org) under the supervision of Prof. Anthony P. Reeves as part of the final project of ECE5470: Computer Vision.