/
Food Manufacturing
Rapid and non-invasive sensory analyses of food products by hyperspectral imaging: Recent application developments

Food Quality Analysis with Hyperspectral Sensors

Hyperspectral imaging is gaining traction as a rapid and non-invasive method for content analysis. As a new tool of food science, hyperspectral can map the polyphenols, including flavonoids, to quantify the flavour profile.

About Opsyne

Opsyne builds and sells OEM hyperspectral imaging technology that can be integrated into complete industrial automation solutions.

Rapid and non-invasive sensory analyses of food products by hyperspectral imaging: Recent application developments

Background

Competition and transparency have created a more demanding and discerning consumer. Consumers prefer food products that meet a certain quality level, particularly sensory properties such as texture and flavour. This is especially important in the alternatives domain, where meat and dairy substitutes strive to be "like" reference animal products. The food industry is seeking technologies and instruments that can correlate consumer preferences with the product's content. This is called sensory analysis.

Hyperspectral imaging is gaining traction as a rapid and non-invasive method for content analysis. As a new tool of food science, hyperspectral can map the polyphenols, including flavonoids, to quantify the flavour profile. This review covers hyperspectral's ability to quantify the sensory properties of food products.

Review

There are four different image acquisition approaches, including point scanning (whiskbroom), line scanning (or push broom), area scanning, and single shot. Line scanning is the most used tool in the food industry as it can scan continuously in one direction, which aligns with the need to fix sensors onto conveyors. The commonly used spectral region is VIS-NIR (400 -1000 nm). The longer wavelengths are more appropriate for, the larger, more complex organic molecules we are looking for in food.

The colour indicates information about the physical, chemical, or microbiological quality of food products. This is of particular interest in meat and fresh produce where colour can indicate shelf life. More accurate predictions of shelf life can aid with sorting to mitigate food waste. Many studies have been carried out to determine the colour features during different processing procedures such as freezing, cooking, pre-cooking, microwave heating, drying, and salting. The colour of the product is one indication of interest.

While examining the defects, chilling injury is one of the most common physiological disorders in fruits and vegetables. It can occur during low-temperature long-term storage and can cause various alterations such as internal browning, deteriorated texture, lack of juiciness, and off-flavour. Hyperspectral in the VIS-NIR spectral range of 400 -1000 nm has been studied to detect chilling injury. 

Bruising is mechanical damage in fruits and vegetables, which can occur during harvesting, packing, and transporting. Researchers have focused on finding automated sorting systems for the early detection of bruises to decrease economic and nutritional value losses. Hyperspectral is used to detect bruises in apples, potatoes, pears, blueberries, peaches and kiwifruits.

The evaluation of tactile sensory features of food products with the help of HSI is of great importance since it replaces the traditional methods such as trained panellists and instrumental procedures. HSI has been explored as a technique to determine the marbling level of meats such as beef and conduct texture profile analysis, scoring the hardness, gumminess, and chewiness of meat products. 

Water holding capacity (WHC) or drip-loss (DL) of meat is identified as the meat's capability to maintain the water inside against external factors such as gravity and temperature. HSI in the spectral range of 400 - 1000 nm and 900 - 2000 nm is a fast and repeatable procedure used to estimate WHC. 

Flavour is a sensory impression of food products, identified by the taste and smell system's combined chemical senses. Fundamentally, there are five taste classes, including sweetness, saltiness, sourness, bitterness, and umami perceived by the mouth. The introduction of HSI for combining spectral and spatial information has had a tremendous impact, such as predicting persimmon's astringency, the phenolic content in red and white grape seeds, the sweetness of melons and tomatoes.

Freshness is a multisensory characteristic. HSI has had a significant breakthrough in the prediction of the freshness of the food. For instance, the freshness of eggs can be quantified using HSI in the spectral range of 400 - 1000 nm. 

Maturity or ripeness is a stage that can show the readiness of foods for eating. The determination of the maturity stage of foods can guarantee sensory quality, leading to market share maintenance by ensuring consumer satisfaction. HSI has been widely investigated to predict and classify maturity stages of different foods such as tomatoes, limes, oranges, cheese, pears, and blueberries.

Implications

Hyperspectral imaging technology has excellent potential as a rapid and non-invasive tool for sensory analysis. Though mostly confined to the research domain, improved data processing and interpretation techniques, including artificial intelligence and machine learning, along with enhanced optics, are opening hyperspectral to becoming a default quality control tool in the food industry. This is especially important in product categories where quality and sensory experience is variable, but consumer expect consistency.