Leaf geometric properties measurement using computer vision system based on camera parameters

Leaf geometric properties play an important role in plat study. This paper aims to propose a method to measure leaf geometric properties, including area, perimeter, length, and width, using a computer vision system. The proposed method measured the properties by capturing leaf image from the top view using a calibrated camera. The captured image was processed to produce a binary image. The properties were extracted from the binary image based on camera parameters. The camera parameters were used to convert the unit of the properties from pixel to cm. Thirty leaf samples from three types of leaf were used to validate the proposed method in an experiment. The experiment result shows that the leaf measurement result using the proposed method has good accuracy with average absolute relative error less than or equal 2.27% and has strong linear relationship with manual measurement indicated by the coefficient of determination greater than 0.999.