TRANSPLANT UNIFORMITY INSPECTION USING MACHINE VISION¹

Uniform flowering is a critical factor in the success of a single truss tomato production system. In this system, tomato plants are grown in crop blocks. Each crop block, contains many plants, upon maturity, preferably be one-pass harvested and rest of the plants will be discarded. One of the ways to improve the uniformity of plant growth is to group plants into different batches base on the external appearance that reflect their internal genetic characteristics. Researchers have found that screened tomato seeds that had similar morphological features grew significantly more uniformly than those unscreened. The morphological features measurements were done manually.

In this study, machine vision algorithms have been developed for the automated measurement of tomato seedling morphological features. Among those, we found the top projected leaf area (TPLA) and the top projected leaf perimeter (TPLP) are most significant in determining seedlings' physiological states.

We have found strong correlations between fresh weight, dry weight and machine vision measured TPLA. The flowering timing of the experimental tomato plants was also found correlated to the machine vision measured TPLA and TPLP. This, in turn, suggests a strong possibility that tomato seedlings can be automatically graded and grouped to improve the physiological uniformity for the tomato production system. With homogeneous plant growing environment, one can eventually improve flowering uniformity.