Original papers
Image processing based acrylamide detection from fried potato chip images using continuous wavelet transform

https://doi.org/10.1016/j.compag.2018.01.012Get rights and content

Highlights

  • Identify potato chips samples containing cancer causing elements, like acrylamide.

  • Segmentation of chips region from background using image processing techniques.

  • Use of Continuous Wavelet Transform for extraction of discriminatory features.

  • Use of statistical features to establish discrimination between potato chips classes.

  • Classifying a test sample as acrylamide or normal using LOOCV based SVM classifier.

  • The proposed method achieved an accuracy of 98.33% with 100% specificity.

Abstract

Fast foods like potato chips and French fries are very common and easily available. The preparation process of such food items is a detrimental factor in deciding if the item is suitable for consumption. Identification of presence of toxic substances like acrylamide in potato chips through conventional methods are time consuming and destructive and needs trained manpower. In the proposed work, an automatic image processing based technique is proposed to detect presence of acrylamide in fried potato chips. The potato chip area is segmented from its background followed by extraction of discriminatory features in the continuous wavelet transform domain using Morlet wavelet. The discriminatory features are analysed strategically and fed to LOOCV based Support Vector Machine classifier to identify presence of acrylamide in the potato chips. The proposed method has an accuracy of 98.33% with 100% specificity. Convincing results and fast computational time indicates that the proposed work can be used for development of non-destructive real-time applications for food quality monitoring.

Introduction

Fast foods like potato chips and French fries are cheaply and easily available in the market and people consume it on the regular basis. They find it tasty and more importantly convenient. These food items are popular among all age groups of people. Although, the preparation of such food items is easy and tastes good, yet the preparation process of these items is one important thing which should be taken care of. The oil and its temperature used for preparation play an important role in keeping these food items healthy and edible (https://nutritionfacts.org/2015/07/21/why-deep-fried-foods-may-cause-cancer/).

When food items, particularly containing carbohydrate, are fried, baked or roasted at the higher temperature (above 120 °C) generates harmful carcinogenic compound (Krishnakumar and Visvanathan, 2014). In potato, this harmful carcinogenic chemical is known as acrylamide. Acrylamide generally does not occur naturally in the potato. It is formed when a reaction of a chemical compound asparagine occurs with reducing sugar. Study based on cancer risk from French Fries has also shown that if a pregnant woman takes such highly deep-fried chip or fries daily then it causes inflammation as well as progression in the cancer (Rommens et al., 2008). Cancer risk level depends upon the percentage of the acrylamide present in the potato chips which in turn depends how long and at what temperature they are fried (Thonning Olesen et al., 2008). Researchers had found out that French fries and potato chip which are baked or fried at higher temperature and for long duration contains acrylamide whose level are more than the considered level of safety to be consumed (Svensson et al., 2003).

Various traditional and conventional laboratory methods exist to detect the presence of acrylamide in fried food items. Such methods involve breaking or damaging of the samples prior to analysis. Examples of such analysis are liquid chromatography (VuralGökmen et al., 2005) and gas chromatography (Reza Ghiasvand, 2016). But these associative analysis takes a lot of time as well as they are very expensive and not easily available in the many food inspection laboratories. It also requires trained manpower to carry on these processes (Pedreschi et al., 2010).

In last few years some work has been reported in non-conventional work for identification of toxic substances from carbohydrate foods and some relevant work are included below.

A colorimetric detection method to detect acrylamide from thermally processed item was proposed by Hu et al. (2016). It can visually detect acrylamide on the basis of Michael addition reaction which is based on nucleophile initiated thiolene. The amount of acrylamide can be determined by visible change in colour which is caused by dispersion/aggregation of gold nanoparticles (AuNPs).

Pedreschi et al. (2006) developed a method to measure color of heterogeneous food items using their shape and color in different color space. Also, they had established a relation between the kinetics of change in color of the potato chips when it is fried at different temperatures. They concluded that change in color of potato chips can be related to formation of acrylamide.

Dutta et al. (2015) proposed a work which involve image processing technique of identification of acrylamide chips from normal chips samples. They had applied discrete wavelet transform domain to analysis the discriminatory features of both classes of chips samples and then applied k-NN classification technique for proper and efficient identification of acrylamide samples. They had also proposed (Dutta et al., 2016) a non-destructive method to find out presence of acrylamide, neurotoxin substance which is common in starch containing food items when they are fired at higher temperature like potato chips. In this work analysis and identification is done in spatial domain. Textural and statistical features are obtained in time domain from the segmented image. SVM classifier is used for identification of acrylamide chip samples and reported an accuracy of 94%.

Yorulmaz (2012) proposed a method for inspection of food that includes the detection acrylamide in cookies when they are fried or cooked at higher temperature and detection of fungal infection in corns. Mottram et al. (2002) had shown that acrylamide is formed in the starchy food items when it is subjected to high heating condition. This occurs due to Maillard reaction between amino acids. Many other researchers also worked on the problem and reported presence on this harmful substance in baked or fried potato food items.

Results of the above-mentioned works are encouraging and may develop interest for many researchers to work more in this field. These works present automatic machine vision which employs non-destructive image processing system for analysis of different classes of potato chips. This might help in increasing the quality of fried food items, which in turn may prove beneficial for people to consume healthy food items and protect them from cancer causing possibilities. Such automated systems have an advantage over conventional methods as these methods provide non-destructive inspection of food item (Brosnan and Sun, 2003). They are cheap and can provide a decision in real time scenarios. Although some work has been already done in this field, yet there is still a need for development of an image processing based real time algorithm which can be helpful in determining the presence of cancer causing chemicals in food items based on visual changes that occur in the samples during preparation.

The main contribution of this work is an automatic method which is a non-destructive and efficient image processing algorithm to identify acrylamide containing potato chips samples. The proposed method employs the segmentation of the area of potato chips from the sample images using an adaptive intensity based threshold to make methodology accurate and invariant of illumination conditions. Wavelet transformed image coefficients are analysed properly and strategically using machine learning. Wavelet domain has with a specific property of performing time-frequency analysis which suits the detection of acrylamide from the images as the occurrence of acrylamide makes discriminatory changes in spatial as well as frequency distribution of the chip image. Discriminatory features are strategically selected and then subjected to classifier based on machine learning. The proposed method has achieved an accuracy of 98.33%. The results are convincing and suggest that the proposed work can be used for development of some real-time application.

The rest of the paper is structured as follows: Section 2 discusses the databases that has been used in the proposed work. Section 3 describes the image processing methods which have been used in the proposed work for the segmentation of the potato chips from the sample image and feature extraction from the segmented image. Section 4 includes experimental results which have been obtained after implementing the proposed method on the dataset of potato chips samples. Section 5 presents the conclusions derived from the results obtained in previous section.

Section snippets

Formation of potato chips samples

For the proposed work, the potato chip samples were prepared in the laboratory. Raw materials required in potato chips preparation are potato and edible vegetable oil. Firstly, the raw potatoes are peeled and washed thoroughly to remove the relatively humidity. Then the washed potatoes are sliced into slices of thickness of 2 mm. Immediately, the sliced potato are rinsed to remove the starch from the surface of sliced potato chips. These sliced potatoes are fried at different temperature and

Preparation of chips samples

Samples of potato chips were prepared according to the procedure prescribed by Gökmen et al. (2007). Small amount of finely grounded samples of potato chips of weight 1 g were obtained by frying them under various temperature and in controlled conditions. Then next step was to suspend 13C3 labelled acrylamide in 10 ml of centrifugal glass tube and 5 ml of methanol. Homogenization was done and this homogenized mixture was centrifuged at 500 rpm for 10 m.100 μl of Carrez-1 and 100 μl of Carrez-2

Proposed methodology

Chemical testing approach is a reliable technique to identify presence of chemicals in food samples. However, it a slow and tedious task and the sample under test becomes non-usable and unfit for consumption, if subjected to chemical testing. So, there arises a need for an alternative approach which can be very useful if there occur some discriminatory visual changes in both sample classes in case of presence or absence of toxic substances. Image processing gives has advantage of performing a

Experimental results

In this proposed algorithm, an image database of 120 samples of potato chips is taken to carry out this experiment. Out of these total images, 60 images are of normal potato chips samples and 60 images are of acrylamide content potato chips samples. Laboratory test i.e. LC–MS test has been done on the same potato chips samples database and labelled them either as normal potato chips sample or acrylamide contained potato chips sample. In the proposed work, the result obtained after applying

Comparison between existing methods and proposed work

Table 5 presents a comparative chart of some existing methods with proposed method. The high accuracy obtained by the proposed method stands in support that the proposed method can play an important contribution in acrylamide detection using image processing. CWT based feature extraction and classification is novel and has never been used before for identification of toxic substances from potato chips sample images. Many of existing methods are either of destructive nature or their results in

Conclusion

This paper proposes an effective non-destructive image processing technique for identification of acrylamide from fried or baked potato chips image. The region of interest is segmented using an intensity based threshold followed by wavelet domain analysis is to extract the discriminatory features from the segmented image of potato chips. The wavelet domain has proved beneficial in differentiating between the two chips classes as the acrylamide containing chips are identified using modulus

References (22)

  • Q. Hu et al.

    A colorimetric detection of acrylamide in potato chips based on nucleophile-initiated thiol–ene Michael addition

    Analyst

    (2016)
  • Cited by (18)

    • Determination of acrylamide in toasts using digital image colorimetry by smartphone

      2022, Food Control
      Citation Excerpt :

      A different colour parameter was used by Dutta et al. (Dutta et al., 2015), making use of grey scale (average values of RGB parameters) as an analytical signal to separate potato chips into two groups, depending on their acrylamide concentration using a support vector machine. More complex approaches were made by the same research unit using neural networks (Singh et al., 2017) and wavelet transforms (Singh et al., 2015; Yadav et al., 2018) with RGB signals as inputs. A multiple linear regression (MLR) using RGB parameters was carried out by Bignardi et al. (Bignardi et al., 2019) in baked potatoes with a R2 coefficient, being G the most representative of the colour change obtained.

    • Intelligent potato frying: Time to say goodbye to the “good old” processing strategies

      2022, Thermal Science and Engineering Progress
      Citation Excerpt :

      Adedipe et al. [3] reported NIR spectroscopy (400–2500 nm) as an efficient way of acrylamide prediction (R-squared = 0.98 and standard error = 135 µg/kg). Instead of direct prediction of acrylamide using spectroscopic methods, many researchers related color changes to acrylamide; the more browning, the higher acrylamide [61,113]. Convolutional Neural Networks (CNN) trained on color images successfully classified acrylamide containing chips with accuracies above 90 % [9,63].

    • An efficient oil content estimation technique using microscopic microalgae images

      2021, Ecological Informatics
      Citation Excerpt :

      If there is an absence of oil-content region then the procedure is terminated else it will proceed further for the oil content analysis through microalgae images. The input digital images are in RGB format which is transformed to HSV (Hue-Saturation-Value) colour-space for proper variation of intensity between the region of interest and its background (Yadav et al., 2018). Contrast enhancement is achieved on the value channel of the images to improve the contrast of the cells, which are present in the microscopic images.

    • A concise review on food quality assessment using digital image processing

      2021, Trends in Food Science and Technology
      Citation Excerpt :

      RMSE values of 1.1 and 1.2 for allura red and 3.3 and 3.0 for tartrazine were reported in case of cross-validation and external validation sets respectively. A fast and automatic method to detect acrylamide in fried potato chips was mentioned by Yadav, Sengar, Issac, and Dutta (2018) using DIP and statistical methods. ROI segmentation was followed by feature extraction by the Morlet wavelet transform, that was fed to LOOCV based SVM classifier to detect acrylamide in the chips.

    • Evaluation of moisture content in processed apple chips using NIRS and wavelength selection techniques

      2019, Infrared Physics and Technology
      Citation Excerpt :

      Process control is of great significance in the food industry, and determining the end of the drying process is critical in commercial food drying. Furthermore, moisture content plays important role in production of fruit and vegetable chips [5,6]. The conventional technique used for determining the end of the drying process is to monitor the moisture content.

    View all citing articles on Scopus
    View full text