The most prevalent method used for tea production makes it difficult to guarantee the tea quality due to the fact that the natural environment is bad and artificial technology level is poor. There is a pressing need to find a feasible way that can estimate tea production schedule and quality. Against this backdrop, this paper introduces the image processing technology into the quality judgment in the tea production to predict chemical constituent changes in tea such that the most appropriate production model will be available. In this study, image processing technology extracts the tea color information in the production process. PCA (principal component analysis) simulates the relationship between the color change information and the chemical composition, based on which to train the BP neural network, and also successfully predict the contents of sugars, starch, total nitrogen, alkalis, and other principal chemical components in the phases of the tea production process. This study provides a reference for tea quality judgment.