A Comprehensive Analysis of the Impact of Illumination Color Variations on the Visual Attributes of Edible Bird Nests

Edible Bird Nest (EBN) products must be processed to ensure cleanliness and quality. The market value of EBN is impacted by color, shape, size, etc. Previous work employed machine vision and machine learning to assist human workers in accelerating the cleaning process. However, illumination is a critical factor influencing the visual attributes extracted from vision systems. However, only a few studies have explored varying color illuminations' influence on these feature attributes. To address this gap, we introduce a framework designed to systematically investigate the effects of various lighting conditions on the extracted features. In this research, a chamber embedded with LEDs with different light colors (white, red, green, and blue) was designed to optimize image acquisition processing by considering the distance and angle of a camera. Then, visual attributes, such as intensity, statistical, and geometry-based features, were extracted and extensively investigated. These extracted features were analyzed using the Analysis of Variance (ANOVA) and Tukey’s tests (Test-1 & Test-2), where the results demonstrated that the images taken under red color were significantly different from images taken under white/blue/green color. Based on the result, red color illumination is suitable for the ML/ DL process, especially for shape classification, but it is also possible with white color illumination. White color illumination is suitable for detecting impurities from EBN because it produces a higher contrast image. The proposed solution has demonstrated great potential in optimizing the lighting condition of machine vision for EBN quality control.