A smart-vision algorithm for counting whiteflies and thrips on sticky traps using two-dimensional Fourier transform spectrum

Although sticky traps are reliable indicators of pest population dynamics but pest counting by humans is time-consuming and menial labour. A novel smart vision algorithm based on two-dimensional Fourier transform (2DFT) spectrum is presented. Rather than directly counting the pests captured on the traps, the novel concept is to treat trapped pests as noise in a two-dimensional (2D) image with 2DFT serving as a specific noise collector. The research objectives included comparing human and 2DFT counting in two proof-of-principle tests: (i) simulated pests with various quantities and distributions arrayed on two series of templates using both ordered and random patterns; (ii) sweet potato whiteflies [Bemisia tabaci (Gennadius), Hemiptera: Aleyrodidae] on yellow sticky traps (YSTs) and western flower thrips [Frankliniella occidentalis (Pergande), Thysanoptera: Thripidae] on blue sticky traps (BSTs). Tests of simulated pests (2-512) on eight templates verified that the 2DFT-based index provides accurate estimates of pests captured on the traps (R2 = 1), independent of pest distribution pattern. High correlations were obtained from count results of whiteflies on 34 YSTs (R2 = 0.9994) and thrips on 33 BSTs (R2 = 0.9989). Measurement errors were addressed.