Abstract
The objective of this study was to establish a standardized color detection method to achieve low-cost, rapid, nonintrusive and accurate characterization of the color change of smoked chicken thighs during the smoking process. This study was based on machine vision technology using the Mean algorithm, K-means algorithm and K-means algorithm + image noise reduction algorithm to establish 3 colorimetric cards for the color of sugar-smoked chicken thighs. The accuracy of the 3 colorimetric cards was verified by the K-medoids algorithm and sensory analysis, respectively. Results showed that all 3 colorimetric cards had significant color gradient changes. From the K-medoids algorithm, the accuracy of the colorimetric card produced by the Mean algorithm, K-means algorithm and K-means algorithm + image noise reduction algorithm was 87.2, 95.1, and 96.7%, respectively. Meanwhile, the verification results of the sensory analysis showed that the accuracy of the Mean algorithm, K-means algorithm and K-means algorithm + image noise reduction algorithm colorimetric card was 69.4, 80.9, and 79.2%, respectively. A comparative analysis found that the colorimetric cards produced by the K-means algorithm and K-means algorithm + image noise reduction have excellent accuracy. These 2 colorimetric cards could become a suitable method for rapid, low-cost, and accurate online color monitoring of smoked chicken.