INGENIEUR
deterioration , disease infection and mechanical injury . As expanding agricultural land is practically impossible in the context of concurrent global warming issues , minimising postharvest losses and increasing agricultural production through an efficient agricultural production system is a sensible strategy .
There are many promising technologies that can be adopted in managing postharvest handling and controlling postharvest losses . The PA in postharvest technology can be divided into two major components , i . e . imaging and spectroscopy . Both technologies rely on the basic principle of the interaction of light with biological systems . Being presumed to be semi-transparent or opaque , agricultural products allow the passage of light at specific wavelengths resulting in the absorption , transmittance and reflectance phenomena . Light reflectance is an intricate phenomenon that can be looked at as specular reflectance , external diffuse reflectance , and scattering . Reflectance conveys information on the structural and chemical characteristics of the product that is very useful for the evaluation of product quality .
Other technologies include an electronic nose ( e-nose ) that mimics the human nose to identify gas odours using chemical sensors , an electronic tongue ( e-tongue ) that measures and compares taste based on dissolved organic and inorganic compound sensors , dielectrics which measure the quality of agricultural products through the interaction of their characteristics with electromagnetic energy , and acoustics that rely on an exciter of an acoustic device which can influence the frequency spectrum to judge the properties and quality of the food products .
These advancements become possible through the integration with the rapid evolution of sophisticated hardware and software . The application of Graphic Processing Units ( GPU ) enables neural nets to facilitate artificial intelligence ( AI ) to learn complex tasks through a massive amount of training data . AI refers to the simulation of human intelligence in machines that are programmed to behave like the human brain and mimic their actions . This is a recent trend that is continuously evolving to benefit not only agriculture but many other industries as well .
Imaging Techniques
Imaging techniques have gained wide attention and interest in recent years and have become an important tool for various activities performed on the production line due to its inherent capability which closely imitates human eyes and brain functions . This technology uses a camera and computer , replacing the human eye and brain function to identify , track , measure and evaluate the target object ( see Figure 1 ). The camera acquires the data and sends a signal to the computer for further analysis and computation for decision making . Among known emerging imaging techniques in postharvest technology are redgreen-blue ( RGB ) imaging , hyperspectral imaging , laser backscattering imaging , ultrasound imaging , magnetic resonance imaging , X-Ray imaging , thermal imaging and fluorescence imaging .
Each technique has its own unique method . RGB imaging for example , is efficient enough to detect the external appearance of an object non-destructively such as colour , shape and size but is unable to provide direct information about the composition or the internal quality due to the limitation of the spectral wavelength of the system . This limitation can be solved by using spectral imaging that can acquire thousands of spectra per sample .
In the “ fingerprint ” of the biological properties of products , the visible light wavelength would be mostly related to the pigment content while the near-infrared wavelength is reported to be correlated to the microstructure and macrostructure of the tissue . Based on the light source and the imaging unit used , the technique can be divided into three categories , namely laser light ( monochromatic ) backscattering imaging ( LLBI ), multispectral backscattering imaging ( MBI ) and hyperspectral backscattering imaging ( HBI ).
In a similar practice to the imaging techniques , the acquired images need to be pre-processed using various algorithms before further analysis by sophisticated machine learning platforms . These emerging imaging techniques have shown promising potential in various conditions of production lines . Hyperspectral imaging has been used for the early detection of bruises in fresh produce . Backscattering imaging has shown promising results for detecting chilling injury in
18 VOL 84 OCTOBER-DECEMBER 2020