Tasting Apples Without Taking a Bite

Hank Hogan

You can't compare apples to oranges. However, people do compare apples to apples, searching for fruit that looks good and that also offers consistent taste and texture.

The problem lies in judging an apple without biting into it. Renfu Lu, an agricultural engineer at the US Department of Agriculture's Agricultural Research Service in East Lansing, Mich., hopes to solve that problem. He is developing a multispectral imaging system that uses several lasers and a camera to nondestructively test an apple. Remote tasting could be integrated into the fruit packing process, allowing suppliers to provide more consistent and higher-quality apples.

Renfu Lu (background) and Benjamin Bailey measure the sugar content of an apple using near-infrared spectra. Photo by Stephen Ausmus.

Lu noted that a variety of methods could be used but that a multispectral approach offers some significant advantages. "We chose multispectral imaging for sensing fruit quality because the technology allows us to quantify light scattering and absorption at selected wavelengths that are important for measuring fruit firmness and soluble solids, or sugar," he explained.

"Light scattering is related to the density and structural characteristics of fruit and thus can be used for measuring fruit texture, or firmness," he added.

The method currently employed by the $2 billion US fruit industry to test firmness is to punch a hole in sample apples using steel probes. Similarly destructive methods are used to determine sweetness. It is presumed that testing a few pieces of fruit establishes the condition of the many others in the batch.

Lu has been conducting research in fruit testing for years. He has picked and photonically sampled hundreds of apples in a project that receives funding from the tree fruit industries of Washington state and Michigan. His prototype optical testing system consists of four diode lasers that were built for this work. One emits in the visible, with a 680-nm center wavelength, while the other three emit in the near-infrared, with center wavelengths of 880, 905 and 940 nm.

The beams from the diodes are combined in the system's optics and bounced off an apple's surface. The reflected light is collected and sent to a near-IR-enhanced CCD camera, which captures the spectral image at discrete wavebands that correspond to the output of the diode lasers.

These reflected intensities are fed into a computer, where specialized algorithms convert the readings into firmness and sweetness measurements. The amount reflected indicates the fruit's firmness, and the amount absorbed tracks the sweetness, or sugar content. The prototype can sort peaches and apples into two or three firmness grades. That, noted Lu, is comparable to or better than the accuracy reported for other nondestructive mechanical techniques.

However, the current prototype is not fast enough to test every apple on the line. A modern packing facility employs conveyor belts or other means such that 10 pieces of fruit pass a given point every second. The prototype can inspect at only one-fifth that rate. Lu said that one of the current research objectives involves improvements so that the system will be faster and easier to calibrate as well as more consistent and reliable.

Another goal is to combine the prototype firmness and sweetness tester with current packing technology, in which machine vision is used for sorting by color and size. "Our technology can be integrated with the machine vision system for inspecting fruit internal quality," Lu said.

He noted that his group would like to work with an equipment manufacturer to transfer the technology -- when it is ready -- to industry. Depending on the cost of lasers and detectors, the final prototype may include more than four wavebands, which could provide an even better measure of fruit quality, he said.

Such commercialization is still a few years off.