Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Identification and Detection for Surface Cracks of Corn Kernel Based on Computer Vision

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2007 ASAE Annual Meeting  073090.(doi:10.13031/2013.23190)
Authors:   Junxiong Zhang, Yi Xun, Wei Li
Keywords:   corn kernel, surface cracks, computer vision, BP neural network

Surface cracks detection of corn kernel has been studied based on BP neural network segmentation of surface color characteristics and morphology algorithm. The seeds of NongDa-4967 and NongDa-3138 (two novel varieties of corn developed by China Agricultural University) were taken as research objectives. Firstly, binary image including the information of cracks, boundary and non-cracks were obtained by horizontal and vertical Sobel operators. Subsequently, by analyzing the color characteristics, a BP neural network model with three layers was built, R, G, B color components were the inputs of the network, and the outputs were background, corn kernel tip cap and other parts. The tip point of the kernel could be identified from the kernel tip cap. And a majority of non-cracks information was eliminated by subtracting a circular area with the tip point as the center. Finally, according to the crack lengths and positions, the crack was extracted, and the lengths were calculated. An experiment has been carried out with 80 kernels with cracks and 80 kernels without cracks selected from NongDa-4967 and NongDa-3138 respectively. The identification results showed that the surface cracks detection of corn kernels could be realized and the detecting accuracy was 92.5% and 88.8%.

(Download PDF)    (Export to EndNotes)