Prism-based multispectral cameras empower high-speed fruit sorting
Over the past few decades, consumer expectations regarding food quality, safety, and freshness have risen steadily. To meet these expectations, fresh fruits and vegetables sold through supermarkets and commercial channels must pass increasingly stringent sorting and inspection procedures.
Post-harvest quality assessment relies on both extrinsic properties, such as color, shape, size, and surface defects, and intrinsic properties, including internal bruising, ripeness, moisture content, sugar level, and firmness. These quality parameters determine the most
suitable downstream application for each product, such as peeling, slicing, frying, canning, or fresh consumption.
Until relatively recently, many of these assessments were performed manually. In several regions, manual inspection is still widely used. However, rising production volumes, labor constraints, and the need for consistent quality across domestic and export markets have made automated optical sorting systems a key enabler for food producers. Automation improves throughput and repeatability, reduces predictive and preventive costs, and increases overall supply chain efficiency.
Visible inspection and its limitations
Standard color( RGB) machine vision cameras are highly effective for identifying surface-level attributes, including visible defects, geometry, color consistency, and surface texture. These systems are widely deployed and remain essential for many sorting tasks.
However, a growing number of quality-related issues are not reliably visible on the surface. Early-stage bruises, internal browning, water-core, moisture gradients, and sugar-related ripeness differences may show little or no visible contrast, especially in the early stages of degradation.
Visible wavelengths( approximately 400 – 700 nm) largely interact with the surface of fruits and vegetables and provide limited penetration below the skin. As a result, inspection systems increasingly combine RGB imaging with nearinfrared( NIR) wavelengths to gain insight into internal product characteristics.
Pear puncture damage separated from texture in the NIR channel
RGB + NIR: combining surface and internal information
By extending inspection into the NIR range( typically ~ 700 – 1000 nm), sorting systems can detect changes related to water content and internal structure that are not visible in RGB alone. Common applications include:
Early-stage blueberry rot detection via the NIR channel
• Detection of internal bruises in apples, pears, and potatoes
• Identification of water-core and internal browning
• Early indicators of texture degradation
Apple stem defect separation from texture in the NIR channel
Using RGB and NIR together allows producers to assess both external appearance and internal condition in a single inspection step, improving throughput and enabling lower system complexity
18 FDPP- www. fdpp. co. uk