Recent Advances in Data Mining Techniques and Their Applications in Hyperspectral Image Processing for the Food Industry

Abstract

Hyperspectral imaging (HSI) facilitates better characterization of intrinsic and extrinsic properties of foods by integrating traditional spectral and image techniques, in which careful and sophisticated data processing plays an important role. In the past decade, much progress has been made on applying various algorithms to deal with hyperspectral images. This review first introduces the general procedure of hyperspectral data analysis and then illustrates the most typically and commonly used algorithms for denoising, feature selection, model establishment, and evaluation, as well as their applications for assessing food quality, safety, and authenticity. Finally, brief summaries for regression and classification methods are presented. This article will provide a guideline for data mining in the future development of HSI in the food field.

Quality Evaluation of Fish and Other Seafood by Traditional and Nondestructive Instrumental Methods: Advantages and Limitations

“…Numerous other mechanical methods have been used to measure texture of fish and other seafoods; nevertheless, there is a little agreement on which is the best method (Cheng et al, 2014). Coppes-Petricorena (2011) summarized various instrumental methods commonly used for determining texture, and indicated that most research studies were carried out using a TA.XT2 texture analyzer.…”

“…One of the main conclusions of this study was that the development of a modified Maxwell model capable of prediction of the fish skin hardness with an error of 0.06%. Some authors were interested in studying changes in fish muscle in terms of texture and structure during postmortem storage and the impact of different handling and processing methods such as ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT 22 smoking, freezing-thawing treatments, and so on (Ayala et al, 2010;Cheng et al, 2014;Sigurgisladottir et al, 2000). Structural changes of sea bream fillet, measured with transmission electron microscopy, exhibited a detachment among fibers up to 5-10 days postmortem as a result of a rapid proteolysis in the muscle tissue, then a loss of I-band, Z line, and actin filaments were observed (Ayala et al, 2010).…”

Selection of feature wavelengths for developing multispectral imaging systems for quality, safety and authenticity of muscle foods-a review

“…The selection of feature wavelengths/variables from hyperspectral data is an interesting task, because the removal of non-informative variables will produce better prediction and result in simpler models (Dai et al, 2014). A good variable selection technique can not only capture variables that are most specifically related to the analyse of interest, but can also exclude regions affected by other sources of variation, leading to the enhancement of the model's robustness (Abrahamsson, Johansson, Sparen, & Lindgren, 2003).…”

“…One way to solve this problem is to decrease the data size to assist in the identification of a few key wavelengths for real-time multispectral imaging implementation (Burger & Gowen, 2011). Key wavelengths may be equally or more efficient than full wavelengths if the selected wavelengths carry most of the spectral information (Dai, Sun, Xiong, Cheng, & Zeng, 2014;. Once the optimal spectral wavelengths are identified, a simple and effective multispectral system can be engineered for industrial applications (Lorente, Aleixos, Gomez-Sanchis, Cubero & Blasco, 2013).…”

“…Extensive reviews have been published on hyperspectral imaging for evaluating food quality and safety (ElMasry, Xiong, Xie, Sun, Zeng, & Liu, 2015). These reviews showed that many research endeavours were concerned especially with key wavelengths selection (Dai et al, 2014;Nakariyakul & Casasent, 2009;Prieto, Roehe, Lavin, Batten, & Andres, 2009;Pu, Sun, Ma, Liu, & Kamruzzaman, 2014;Zou, Zhao, Povey, Holmes, & Mao, 2010), but a comprehensive review on selection of feature wavelengths associated with specific characteristics of the quality and safety parameters for particular application is not available. The availability of a large and increasing number of studies with continuous emergence of new wavelength selection techniques in recent years necessitates such a review, which will be highly valuable to develop simple and cost effective multispectral sensors for a particular application.…”