The colour and shape of food are two highly valued qualities. But nowhere are they more important than in the fast-food industry, where consumers demand a uniform quality of product each time they visit a restaurant, writes Dave Wilson.

Unsurprisingly, the bakeries that supply the buns and rolls to customers such as Burger King and McDonald’s are always looking for ways to improve the measures by which they can quantify the quality of bun size and colour.

Current means to inspect such products are labour intensive. Examining a product for shape and colour correctness, for example, involves the removal of a few samples of the product each hour from a production line and the manual inspection of the bun against customer specifications.

In many cases, the colour matching procedure is carried out using a baking colorimeter a handheld device that can measure lightness and darkness, reporting results in the commonly used CIE L*a*b* colour space and scale.

But despite their effectiveness, the accurate control of the colour quality of the buns coming out of the ovens is still a challenge, partly owing to production rates that can see buns produced at rates as high as 1,000 per minute.

In most large bakeries, once the buns leave the oven, they are transferred to cooling racks and then to a bagging area to be packed. Only once they arrive there are they checked for colour uniformity and shape. At this late stage, however, any colour imperfections that are identified could potentially mean that hours of unusable products may have already been cooked and racked before the parameters of the baking oven can be adjusted to redress the issue.

Obviously, it would be beneficial to both the bakery and its customers if a more accurate assessment process could be deployed, enabling the oven parameters to be controlled through the use of an automated system that could correct the colour of the products produced in them long before it drifts out of specification.

Now, researchers at Georgia Tech, led by Dr Doug Britton, have designed a supervisory oven control system that can do just that capturing colour images of the buns and using this data to control the temperature of the ovens the buns are cooked in.

Since developing the system, the research team has licensed it to Baketech, a developer of process monitoring systems that can be either added to existing bakery lines or incorporated into new lines. And now, Baketech has installed a first prototype of the system at a plant owned by the Flowers Baking Company, which has several baking facilities across Georgia.

To identify potential defects, the system employs high luminescent stroboscopic LED strobe lights, which illuminate the buns as they leave the oven, and a pair of cameras that then capture images of them. Both cameras and the light sources sit just 2ft over the belt carrying the buns out of the oven and are controlled by a personal computer sitting in an independent rack on the production floor.

The images of the buns captured by the cameras are transferred to the personal computer, where proprietary image processing algorithms calculate the shape and the colour of the buns in the CIE L*a*b* colour space matching the existing colour manual inspection standards already in place at the bakery. The colour data is then sent to an existing controller on the oven that uses it to adjust the temperature of the burners inside it.

Improving the humidity response could allow us to better control the colour, leading to tighter tolerances

To keep the bun colour within acceptable limits, Britton said it was important to ensure that the control feedback software only corrected the temperature of the oven within a specific range of accepted temperature margins. This was to ensure it could not over or under-correct for any inconsistencies in the products. The team also had to consider the large thermal mass of the huge ovens and the time lag that takes place between updating the controller oven settings and the time the oven would take to meet any new set-point temperatures.

According to Britton, the team could also control the speed of the line to achieve the same goals. But it found that doing so could cause problems with cooking schedules and lead to backlogs. The team also investigated the possibility of controlling humidity, but the humidity control was too slow to respond to make a change in the product colour.

’Nevertheless, improving the humidity response could allow us to control the colour even better, leading to even tighter colour tolerances, which is something that is being investigated,’ said Britton.

From an image processing perspective, the challenge was to develop a system that could capture uniform images that would enable feature extraction to be reliably performed on a continuous basis in a bakery environment right at the exit of an oven, where temperatures can be as high as 400º.

To perform this function, it was necessary to cool the cameras so that uniform imaging could be performed regardless of temperature. The personal computer that ran the image processing application needed to be cooled too, because even the ambient temperature in the bakery in summer can be as high as 110º.

The developers at Georgia Tech also realised that, ideally, they would like to enable their system to be retrofitted to almost any conveying system in a bakery. For this reason, they developed the image processing algorithms so they would work equally well on systems that moved product on wire belts and solid coloured belts, as well as being capable of imaging the buns before they were removed from their baking pans.

Aside from extracting data on the colour and shape of the buns, they recognized they could also capture additional data relating to bun size, baking residue and any foreign material on the bun, as well as the coverage of toppings such as sesame seeds.

Although this information is not used per se by the supervisory oven control system to control the temperature of the oven, it can be correlated to the other parameters that are, with the result that production personnel at a bakery have now been provided with many additional measures of quality control.

’We are logging all the process characteristics relating to each batch of buns that is cooked the line speed as well as the cook time and both the set-point temperature and the actual temperature of the oven,’ said Britton. ’This allows production engineers to use the historical data as a tool to help them trace the potential causes of any issues in production.’

He added that the new system could be further improved so the image data that is captured could be used to control a pass/fail system that would be automatically capable of rejecting defective buns from the line from the moment they emerge from the oven.

Britton claimed that the system could be adapted for use in bakery production lines making biscuits, cookies, crackers, bread and pies. ’The food industry is extremely interested in ensuring the uniformity of the colour of the products that are produced from bakeries, so systems such as ours that can improve the consistency of the colour of their buns will be a huge benefit,’ he said.

Source: source.theengineer.co.uk