Parametric Detection of Rice Kernel Shape Using Machine Vision
Detecting broken kernels is a fundamental task in rice quality assessment. This research inves-tigated an automated parametric detection method for detecting rice kernel shape through deter-mining a set of kernel shape parameters in a real-time image process procedure. The core of this procedure included the separation of touching rice kernels and the description of rice shape using a set of defined parameters. The touched kernels separation method first identifies touching ker-nels using an image regional area threshold, then searches for separation zone according to the concavo convex of kernels image boundary. An approach of eight boundary feature points identi-fication method was developed to describe the shape of rice kernels, and then to distinguish the broken ones in terms of the obtained values of those eight boundary feature parameters. A valida-tion test verified the accurate rate of this developed method was over 86％ in a real-time rice quality assessment.
Machine Vision Rice Kernel Shape Detection Touched Kernel Separation Shape Feature Extraction
Yaqin Wang Hua Gao Yong Liang
College of Information Science and Engineering, Shandong Agricultural University, Taian, Shandong 271018, P. R. China
国际会议
南昌
英文
1212-1219
2010-10-22（万方平台首次上网日期，不代表论文的发表时间）