CC BY
18
Научни трудове на Съюза на учените в България-Пловдив. Серия В. Техника и технологии, т. XV, ISSN 1311 -9419 (Print), ISSN 2534-9384 (On- line), 2017. Scientific Works of the Union of Scientists in Bulgaria-Plovdiv, series C. Technics and Technologies, Vol. XV., ISSN 1311 -9419 (Print), ISSN 2534-9384 (On- line), 2017.
УСЪВЪРШЕНСТВАНЕ ОБЛАСТТА НА ИНТЕРЕС ПРИ ОПРЕДЕЛЯНЕ ШУПЛИВОСТТА НА ТИПОВХЛЯБ ЧРЕЗ ОБРАБОТКА НАИНАТРАЖЕН1М Ангел М. Данев1, Христинн Аедтсева1, АтаоасхтАос акова-Арденска1,
ЛенаКостадинова-Лгоргиева1 1Катедра Компютьрни вистеми а теонолтгиа, Технически факултет, Униве^итет но храеателни ^^^нототи^^, Пловдив, Балгария
REGION OF INTEREST IMPROVEMENT BY IMAGE PROCESSING IN BROWN BRERO POROSITY EVALUATION Angel M. Danev1, Hristina An^eeva1, AtanasLa Bosakwa-LrOenska1,
Lena KostaOinova-Georniena1 1Department of Computer Systems and Technoloaies, Technical Faculty, University of Food Technologies, Plovgiv, Bu^ania
Abstract: This paper presents a research over using of different shapes for ROI (region of interests) definition by image processing of brown bread. Two types of analyzes are made -physicochemical and computer one (through image processing) for evaluation of porosity of the middle of bread. A software product with GUI (Graphical User Interface) implemented in C# is developed for brown bread images binarization. Two types of shapes for ROI are applied - rectangular and elliptical one. A calculation of the percentage of white pixels to the total number of pixels in the images using both types of ROI is made. These values correspond with the porosity of the middle obtained by physicochemical method.
Key words: region of interest (ROI), image processing, brown bread, bread porosity, C#
Introduction
The brown bread takes an important place in the food industry. It is produced by approved standard "Bulgaria" according to which the porosity of the middle of brown bread should not be less than 65%. The main raw material for the brown bread production is wheat flour. The quality of the finished product depends on the used raw materials - wheat flour, bread yeast, iodized cooking salt, technological additives such as sugars, fats and so on [Approved Standard "BULGARIA", 2011]. For evaluating the quality of the bread the following parameters are analyzed - organoleptic, physicochemical, existence of chemical impurities and microbiological ones. In this paper a contactless and a non-destructive method for evaluating the physicochemical parameter such as the brown bread porosity of the middle is presented.
Materials
Four different brands of brown bread have been purchased from a marketplace with the aim of accomplishing the analysis. The bread is produced by approved standard "Bulgaria". For
physicochemical analysis implementation a measuring cylinder is used. It has measurement scale up to 100ml. At each measurement, the cylinder is filled with water exactly 50ml. From the middle of each slice a rectangle with size 3 by 6 cm. is cut. The taken sample is cut again to 3 equal in size pieces. Each of the pieces is creasing in the shape of a small ball. Every single ball is dropped into the measurement cylinder and the volume Vi [cm3] is evaluated. The overall volume V of the tested sample is a sum of the volumes of all three balls. The porosity of the bread is calculated using the following formula:
rV - Vn
X =
100
V
To determine the porosity of the brown bread using the modem computer systems, a special software product is developed. It is used for digital processing of images captured by camera. A special region of interest is chosen and the porosity of the middle is calculated algorithmically. For getting images a digital camera Olimpus PEN mini E-PMI is used. The technical characteristics of the camera are the following:
- resolution: 12.3Mp;
- type of the optical sensor: Live MOS;
- processor: TruePic VI;
- image processing format: JPEG, MPO, RAW + JPEG;
- lenses: Micro Four Thirds;
- minimum focal length: 9.8 inches;
A tests of the algorithms and analyzes of the experimental data are made by using a modern computer system.
Methods
In this paper a comparison between two methods for analyzing and control of brown bread porosity is made. The physicochemical method has two significant disadvantages. On the one hand, the time needed for processing physicochemical analysis of brown bread porosity is too long, on the other hand this method is contact and brake the integrity of the sample [Andreeva H., Kostadinova-Georgieva L., Bosakova-Ardenska A., 2013]. Unlike the physicochemical method, the computer one provides an opportunity for making much faster analyzes. This method is not contact and it doesn't broke the integrity of the sample. The experimental setting shown in fig. 1 is used for capturing each slice from both sides. It is composed of following elements: 1) tripod for locking the camera in stable position; 2) lever for precise positioning the camera; 3) digital camera - OLIMPUS PEN mini E-PMI; 4) the analyzed slice of bread; 5) a pad with black color (it is used for unification the background color of the image).
The captured images are loaded in the software product and a region of interest (ROI) is defined. A ROI with elliptical shape is chosen. This region of interest is cut from the image and it is saved in bmp format. The new image (containing the ROI) is loaded in to the computer program and after that it is binarized. The received binary image is processed by the Tsai algorithm. The porosity of the brown bread is calculated on that base of percentage ratio of white to total number of all pixels in the image [Andreeva H., Kostadinova-Georgieva L., Bosakova-Ardenska A., 2014].
Fig. 1. The experimental setting
The obtained values correspond to the porosity of the brown bread received from the physicochemical method.
Computer software for analyzing the porosity of the middle of the bread
Software implementation
It is used the high-level programing language C# for developing the computer program [Nakov S., Kolev V., 2011]. The graphic user interface and the main screen of the program is shown on fig. 2. When the user starts the program a dialog window appears on the screen. Then
using a dropdown menu the user can choose between two types of analyzes. The first one is called Tsai moment preserving thresholding [Tsai, W-H, 1985] and the second one is called Vector median thresholding [Andreeva H., Kostadinova-Georgieva L., Bosakova-Ardenska A., 2013]. After pressing the Start Experiment button the main window of the program appears. In the first step, the original image of the whole slice is loading. After pressing the Save Cropped Image button a new image containing only the ROI that we choose is saved. For the purpose of this study a ROI with elliptical shape is chosen.
Fig. 2. The GUI of the computer program. Loaded original image and the binarized image obtained after Tsai processing.
Regions of interest (ROIs) definition and improvement
To determine the exact region of interest on the image have been taken into consideration the following parameters - the distance between the surface and the camera lens; the image size in pixels; the corresponding size of the captured image in centimeters. Fig. 4 illustrates example of an image used in a previous study in which a ROI with rectangular shape is used. On the other hand, fig. 3 a) shows an image with the elliptical ROI. After processing that image with the Tsai algorithm the image shown in fig 3. b) is obtained. Using an image with elliptical ROI is expected to achieve greater accuracy in obtaining the final results. The rectangular ROI contains less pixels than the elliptical ROI.
Fig. 3 a) Cropped elliptical ROI Fig. 3 b) Binarized image after ROI Fig. 4 Cropped rectangular
Tsai processing
Results
In fig. 5 the accuracy of the methods used for evaluating the brown bread porosity of the middle is illustrated graphically. All of the data corresponding to the porosity of the middle are plotted on the graphic. Also the minimum allowable value (shown with straight line) for the brown bread porosity of the middle, according to the established "Bulgaria" standard is illustrated on the graphic. According to the examinations that have been made and the results that have been obtained it can be concluded that all of the analyzed breads meet the requirements of this standard. The values obtained after the analysis are shown in table 1.
Table 1. Results for brown bread porosity of the middle in percentages obtained using elliptical and rectangular ROI in comparison with physicochemical method_
Bread 1 Bread 2 Bread 3 Bread 4
Rectangular ROI , X[%] 96,22 96,67 97,25 96,21
Elliptical ROI , X[%] 96,44 96,69 97,05 96,26
Physicochemical method, X[%] 84,90 88,20 88,70 85,00
Brown Bread Porosity of the middle by image processing [%]
90
so
70
60
Bread 1 Bread 2 Bread 3 Bread 4
-•-Elliptical ROI -»-Rectangular ROI -•— Physicochemical method -»-Standart "Bulgaria"
Fig. 5. Graphical representation of the results for brown bread porosity using elliptical ROI
Table 2. Correlation analysis
Elliptical ROI Rectangular ROI Physicochemical method
Elliptical ROI 1
Rectangular ROI 0,975 1
Physicochemical method 0,911 0,920 1
For comparison between the two methods presented above a correlation analysis is performed. From the values obtained after correlation analysis shown in table 2 it can be made the conclusion that the correlation is very high between physicochemical method and the elliptical/rectangular ROI.
Conclusion
According to the results received from the analyzes that have been made it can be made a conclusion that the method using computer software with algorithms for digital image processing with elliptical ROI in term of evaluating the brown bread porosity of the middle can provide results close to the reference method such as physicochemical one.
References
Approved Standard "BULGARIA" - 04/2011 - Bread "Brown", Ministry of Agriculture and Food, Sofia, 2011 (in Bulgarian)
H. Andreeva, L. Kostadinova-Georgieva, A. Bosakova-Ardenska, Evaluating the porosity of bread by image processing, Food science, technology and technology (in Bulgarian), 2013, Scientific works of University of food technologies, Volume LX, ISSN 1314-7102, pp 1136-1139 S.Nakov, V.Kolev, Introduction to programing with C# (in bulgarian), Veliko Turnovo, 2011, ISBN: 978-954-400-527-6, pp. 23 -25
H. Andreeva, A. Bosakova-Ardenska, L. Kostadinova-Georgieva, Comparison of two methods for automatic image binarization, Scientific conference "Computer science and technologies", Varna, 2014, ISSN 1312-3335, pp 138-143
Tsai, W-H, Moment-Preserving Thresholding: A New Approach, in CVGIP, 1985, vol.29, pp 377-393
