A Computer Vision System for the Assessment of Ice Cream Melting Behavior

Here, we present a protocol based on a computer vision system (CVS) to determine the melting behavior of multi-phase food systems.

This research aims to propose a computer vision system procedure to study one of the most important features of ice cream, which is a melting behavior. Usually, melting behavior is studied by a gravimetric approach, giving information about the initial time of melting and the rate of melting. However, the aspect of the product during melting is also very important, and this method calculates a couple of indexes related to the shape and size retention of ice cream during the meltdown.

The advantage of using a computer vision system in the ice cream melting evaluation, relies on the increased sensitivity in detecting small variations, compared to the gravimetric method and on the possibility to elaborate data related to the visual appearance of the product during meltdown. The computer vision system approach here proposed can be easily applied to other food products for which the melting rate and shape retention index are important, such as morsels and forms like whipping milk cream and egg albumin. These can lead to a better understanding of the meltdown mechanism, which is related to different factors.

To begin, select transparent cups with fixed volume and shape, each equipped with a lid. Cut two strips of baking paper, each about two centimeters wide. Using paper tape, attach these strips inside the cup walls, forming across at the bottom.

Fill the cup with ice cream using a spatula. Once full, smooth the surface of the ice cream with the spatula and seal the cup with its lid. Store the prepared sample at minus 30 degrees Celsius for at least 24 hours, then condition the sample at minus 16 degrees Celsius for 24 hours before analysis.

For the melting trial, set a thermostatic cabinet to approximately 20 degrees Celsius. Place a digital scale inside the cabinet. Connect the scale to a computer with software to record weight as a function of time.

Next position a graduated cylinder on the scale and reset the weight. Place a hanging funnel over the cylinder to collect melted ice cream. Set up a camera on a tripod in front of the cabinet, ensuring the best framing for sample images.

Then prepare a metal wire mesh screen with a size reference. Remove the ice cream cup from the freezer and detach the ice cream from the cup walls, using a spatula, keeping the baking paper attached. Gently place it on the wire mesh screen.

Then remove the baking paper strips. Position the screen with the ice cream on the funnel inside the cabinet. Capture the initial image with the open cabinet door, including the size reference.

Record data for 90 minutes and take pictures of the ice cream every 15 minutes. Using this protocol, melting trials were performed for the three ice cream samples made with different sweeteners such as sucrose, sucromalt, and erythritol. While all samples showed good shape retention after 90 minutes, the lowest melting rate was observed for the ice cream made with erythritol.

To begin, download the images of melting ice cream captured by the digital camera setup to a computer. Then in the image analysis software, click on file and open, to open the ice cream images. Start processing the first image and click on edit and rotate to adjust its rotation if necessary.

Then zoom into the size reference and click on measure, then calibration, followed by spatial. Next, click on image to spatially calibrate the image. Move the scale over the size reference and calibrate the length to convert pixels to millimeters.

Next, click edit new AOI and rectangular, to select a rectangular area of interest that includes the ice cream sample, avoiding the edge of the metal wire mesh screen. Crop this area and convert it to gray scale by clicking edit, then convert to and selecting gray scale. To apply the best fit filter, click on enhance, then equalize, and best fit to automatically adjust brightness and contrast.

Then open the measure and count size window and set the parameters of area, box height, and box width. Choose manual measurement in the count size window and click on select ranges to set the histogram values to segment the bright shape of the ice cream. After closing the segmentation window, click on count in the count size window of the software to measure the parameters.

View the results by clicking on view and measurement data. Then click on file and data to clipboard to copy these results and paste them into a data management software spreadsheet. To evaluate the melting index, use this equation on the H and W data at different melting times, to calculate the shape retention index.

Refer R and A data at each time point to the corresponding index at time zero, and plot these results over time according to these equations. For gravimetric data analysis, save the spreadsheet with the weight of the melted sample per minute. At the end of the 90 minute melting trial, open this spreadsheet in data management software to create a melting curve showing the melted weight over time.

Select data from the linear portion of the melting curve to calculate the least squares regression line and the regression coefficient. The slope of this line represents the ice cream melting rate in grams per minute. Calculate the starting time of melting as the X intercept, where Y is zero.

Using this protocol, melting behavior was analyzed for ice creams made with different sweeteners such as sucrose, sucromalt, and erythritol. Shape retention curves showed that ice cream made with sucrose lost its shape faster than the others, whereas the ice cream made with erythritol kept its shape until the end of the trial. Similarly, area retention curves showed that ice cream containing erythritol showed a lower meltdown rate compared to ice creams with sucromalt and sucrose.

Further, gravimetric analysis showed that the ice cream containing erythritol showed the highest melting starting time and lowest melting rate out of the three samples.