Feasibility Study of silkworm pupa sex identification with pattern matching

Abstract

Silkworm sex identification is one of the important processes in the sericulture industry because it can assist in effectively separating strong and healthy silkworm pupae from the weak ones. In this paper, we study and show that a desired moderate accuracy and fast response time in silkworm gender identification can be realized by deploying a widely-used and simple normalized cross correlation (NCC)-based pattern matching operation in our optical penetration-based silkworm pupa gender identification structure. Other key features are ease of implementation, adaptive learning ability, and low component counts. Experimental proof of concept is performed by using three 636-nm wavelength light emitting diodes, one 1600×1200-pixel web camera, an 8-bit microcontroller, a notebook computer, and our LabView program. There are 25 female and 20 male silkworm pupae under the study. Experimental results show that male silkworm pupae can be completely identified under the NCC-based pattern matching performed in both the Cartesian and polar coordinate systems. For the female silkworm pupa, the best measured accuracies of 80% and 56% are obtained by performing the NCC-based pattern matching operations in the Cartesian and polar coordinate systems, respectively. The NCC-based pattern matching operation in the Cartesian coordinate system also offers a measured 13-ms response time, which is twice faster than in the polar coordinate system. The moderate measured total accuracy of 88% and fast silkworm gender identification shows high potential for the deployment in the sericulture industry.