Determine the color change of fresh green lettuce by using reflectance reconstruction from RGB image

Abstract

This paper presents an image processing technique to determine the color change of salad lettuce is stored at 15ºC for storage times of 0 to 5 days. The technique divided the color of salad lettuce into 8 clusters (Dark-green, Light-green, Green-yellow, Brown, Dark, White, Shadow, and background) and used these clusters for spatial and spectral analysis. In the case of spatial analysis, the number of pixels of each cluster was countering over storage time for calculating the area of each cluster in the image and was used to determine the color change of the lettuce salad. In cases of spectral analysis, the reflectance reconstruction technique was applied to reconstruct the reflectance data from the image. RGB values from these images were transformed to tri-stimulus values (XYZ) and L*a*b* and then used with a trust-region-dogleg algorithm for reconstruction the reflectance from L*a*b* values. The reflectance data were normalized by an average sum of reflectance and called relative reflectance, and then use in the partial relative reflectance in a range of blue (450-500 nm), green (500-570 nm), and red (610-650 nm) to calculate the spectral gradient. The spectral gradient was used to determine the color change of the lettuce salad over storage. The result of both spatial and spectral analysis shows that changes in the colors of lettuce can be detected at storage time in days 3.

Keywords

• lettuce
• image processing
• reconstruction
• color change