Harness machine vision to improve produce quality testing, inspection.

Currently, batches of fruit and vegetables are judged by sample tastings. Samples are also tested for firmness by mechanically stabbing them with a thick steel probe. With both methods, the tested produce has to be thrown away. Moreover, there is no guarantee that all of the produce in the batch will taste the same.

Now, USDA-ARS scientists are optimizing optical sensors that can predict the quality of fruit or vegetable flavor--right after picking and in the packing plant--nondestructively. Nondestructive technologies for grading and sorting fruit by internal quality, such as firmness and sugar and acid content, would ensure a consistent premium quality product, increase consumer satisfaction and enhance the fruit industry's competitiveness and profitability.

Scientists with the ARS Sugarbeet and Bean Research Unit, East Lansing, MI, sample apples with a prototype optical detector. The detector fuses four laser beams, each at a different waveband of light, into one. Light photons momentarily scatter all the way to a fruit's core. An imaging spectrograph, a digital camera and software analyze the amount of laser light absorbed by the apples, which indicates sweetness. The amount of light bounced back after interacting with fruit tissue reflects fruit firmness. Sweet and sour tastes are a factor for apples, cherries, peaches, and other fruit, but firmness analysis is often more important to consumers and is technologically more challenging.

Multispectral imaging combines spectroscopy, which analyzes light wavelengths, with machine vision, which enables a computer to "see." When commercialized, the optical sensor would be used by the produce industry to sort fruit just after it's been picked. Researchers also built a larger version fitted into a mini-packing line for lab use. It's a prototype for a machine that would be used on fruit-processing lines to make a second quality check after some time had passed and the fruit had been handled.

The sensors can sort peaches and apples into two or three firmness grades. The technology is relatively easy to implement for rapid online sorting and has potential for measuring multiple quality attributes simultaneously. The sensors work better on peaches than on apples in terms of firmness measurements. Apples are challenging because they are more variable in firmness and have a narrower firmness range from apple to apple. The sugar-content predictions for apples compared well with actual sugar-content measurements.

The goal of scientists is to sign a cooperative research and development agreement with companies to commercialize prototypes for use in fruit quality labs, packinghouses and orchards. They are working to speed up scanning speed to match that of commercial apple conveyors: 10 fruits per second. The ability of a spectrograph to capture images from four light bands at once makes this speed possible.

The researchers considered patenting the technology. But they decided not to and instead will consider working with an equipment manufacturer through a cooperative agreement once the technology is ready for commercial use. They would like to transfer the technology to industry over the next two to three years. A system would not cost much more than the machine vision systems currently used in many fruit processing plants.

Further information. Renfu Lu, USDA-ARS Sugarbeet and Bean Research Unit, 224 Farrall Hall, Michigan State University, East Lansing, MI 48824; phone: [phone]; fax: [phone]; email: [e-mail].