Submitted to: ASABE Annual International Meeting
Publication Type: Abstract only
Publication Acceptance Date: 9/1/2011
Publication Date: 9/1/2011
Citation: Yang, C., Kim, M.S., Chao, K. 2011. The application of hypserspectral imaging analysis to fresh food safety inspection. ASABE Annual International Meeting. Interpretive Summary:
Technical Abstract: Line-scan hyperspectral images of fresh matured tomatoes were collected for image analysis. Algorithms were developed, based on spectral analysis, to detect defect of cracks on fresh produce. Four wavebands of 569 nm, 645 nm, 702 nm and 887 nm were selected from spectra analysis to use the relative intensities at these wavebands in two spectral ratios. A simple threshold line was derived using these two ratios to successfully differentiate normal fresh matured tomatoes from cracked ones. Because the algorithms were developed for potential implementation in a machine vision system on food processing lines, the algorithms were intentionally designed to be simple with only a few certain wavebands required, for quick data transfer from the EMCCD camera to computer and for fast computation. The spectral machine vision system in which the two-ratio algorithm can be implemented is a system capable of either continuous (broadband) hyperspectral imaging or discrete multispectral imaging and of easily switching between these two modes of operation. Implementation of the algorithm could allow for precise recognition and differentiation of normal red tomatoes from defect cracked tomatoes to ensure food quality and safety. The line-scan imaging system with the proposed algorithm could help to remove potentially infected fresh tomatoes from processing lines before shipping to market.