sciepub.com
Quick Submission
Using the Properties of Wavelet Coefficients of Time Series for Image Analysis and Processing
Vyacheslav V. Lyashenko1, Rami Matarneh2,
, Zhanna V. Deineko3
1Department of Informatics, Kharkov National University of Radio Electronics, Kharkov, Ukraine
2Department of Computer Science Prince Sattam Bin Abdulaziz University Saudi Arabia, Al-Kharj
3Department of Media Systems and Technology, Kharkov National University of Radio Electronics, Kharkov, Ukraine
Abstract
Image processing is used in many fields of knowledge; because it allows to automate processes to get more information about the object being studied. Image processing techniques are many and varied. Wavelet analysis is one of such techniques. Among various methods and approaches of wavelet processing we distinguish the ideology of multiresolution wavelet analysis. The essence of this ideology is to perform wavelet decomposition on test data and the subsequent analysis of the relevant factors of this decomposition (the wavelet coefficients). An important aspect is the consideration of the properties of the wavelet coefficients. Based on this, we have examined the feasibility of using the properties of detailing wavelet coefficients to study and compare different images. We have introduced additional characteristics of images on the basis of sets of detailing wavelet coefficients decomposition. These characteristics reflect the dynamics of change in mean and variance for the detailing of the coefficients of the wavelet decomposition. We have shown that the dynamic changes in mean and variance of detailing coefficients of wavelet decomposition can be used to analyze and compare different images.
Keywords: image, wavelet transform, image processing, wavelet coefficients, detailing coefficients, approximating coefficients, levels of wavelet decomposition
Copyright © 2016 Science and Education Publishing. All Rights Reserved.Cite this article:
- Vyacheslav V. Lyashenko, Rami Matarneh, Zhanna V. Deineko. Using the Properties of Wavelet Coefficients of Time Series for Image Analysis and Processing. Journal of Computer Sciences and Applications. Vol. 4, No. 2, 2016, pp 27-34. https://pubs.sciepub.com/jcsa/4/2/1
- Lyashenko, Vyacheslav V., Rami Matarneh, and Zhanna V. Deineko. "Using the Properties of Wavelet Coefficients of Time Series for Image Analysis and Processing." Journal of Computer Sciences and Applications 4.2 (2016): 27-34.
- Lyashenko, V. V. , Matarneh, R. , & Deineko, Z. V. (2016). Using the Properties of Wavelet Coefficients of Time Series for Image Analysis and Processing. Journal of Computer Sciences and Applications, 4(2), 27-34.
- Lyashenko, Vyacheslav V., Rami Matarneh, and Zhanna V. Deineko. "Using the Properties of Wavelet Coefficients of Time Series for Image Analysis and Processing." Journal of Computer Sciences and Applications 4, no. 2 (2016): 27-34.
| Import into BibTeX | Import into EndNote | Import into RefMan | Import into RefWorks |
At a glance: Figures
1. Introduction
Image analysis is a powerful tool in various fields of knowledge; because of significant part of information that can be received. At the same time, various methods of processing and interpretation of images are form the basis of some of artificial intelligent systems. If we take into consideration the possibility of applying various methods of image analysis and processing in a combined form with an interconnected methodology, this inevitably will play an important role to solve practical problems.
On the other hand, if we take into consideration the possibility of applying various methods of image analysis and processing in a separate form, this will allow us to: (1) choose the most comprehensive methods of image analysis and processing, (2) to optimize the structure of artificial intelligent systems which are based on the analysis of images and (3) to increase the productivity and the overall performance of data analysis. Usually, images of the real world are presented in a two dimensional form, which allows the use of separate methods of image analysis and processing. These methods can be classified into three category: (1) Methods of preliminary image processing (noise suppression, contrast increase, localization of separate sites of the image) [1, 2], (2) Methods of preliminary analysis (segmentation, contour allocation) [3, 4], (3) Cognitive recognition methods [5, 6, 7].
The mentioned methods use different mathematical apparatus: group theory, linear algebra, graph theory. Currently, the image processing technique is widely use wavelet analysis [8, 9, 10]. This is due to the fact that the use of wavelet analysis procedure yields: the possibility of processing for various aspects of the studied images from the point of view of their method of presentation and the ability to use different approaches to highlight any aspects presentation of visual images during subsequent processing. These reasons determines the basis of our research idea.
2. Materials and Methods
2.1. Wavelet Analysis as a Tool for Image ProcessingThe main idea of wavelet transform is a time-and-frequency signal notation [11]. Wavelet transform is a decomposition with the use of functions, each of which is a shifted and scaled copy of one function – mother wavelet [11, 12]. Wavelet in this case is a function rapidly decreasing to infinity with average value equals to zero. If the signal is discontinuous, only those wavelets will have high amplitudes, where the maximum value will appear near the discontinuity point. At the same time, discontinuity point is a sharp intermittent transition during some process. Quantitatively, it can be estimated by the value of the first derivative of such process, taking into consideration that the first derivative of intermittent transitions is very high [10, 12]. The real processes in reality cannot have perfect discontinuity points. In fact, the measured fractal transitions are characterized by the finite value of the derivative. The sharper the transition, the higher the derivative value is. Smooth transitions will have small derivative values. This allows us to determine the presence of special characteristics of the analyzed image, as well as the point where these characteristics may arise [10, 12]. At the same time we can present two-dimensional image as a vector (Figure 1).
Download as
Then, for the image analysis, we can use the so-called multiresolution analysis method, on the basis of the theory of wavelet transformations [13, 14, 15]. A multiresolution wavelet-analysis transforms time series to hierarchical structure by means of the wavelet transformations which results to the set of wavelet coefficients. On each new level of wavelet-expansion there is a division of an approximating signal of the previous level of detail (presented by some time series) on its high-frequency component and on more smoothed approximating signal [13, 15]. According to discrete wavelet-transformation time series
, (
, 
) consists of a set of coefficients – detailing and approximating [16]:
 | (1) |
Where 
– Approximating wavelet-coefficients of level 
;
– Detailing wavelet-coefficients of level
;
– Chosen maximum level of expansion;
– Quantity of detailing coefficients at 
level of expansion;
– Quantity of approximating coefficients at level
– Mother wavelet-function
– Corresponding scaling-function.
At set the mother wavelet and 
corresponding scaling-function approximating 
coefficients and 
detailing coefficients of 
DWT (discrete wavelet transformation) for the process 
can be defined as follows [16]:
 | (2) |
 | (3) |
In particular, on each level of discrete wavelet transformation detailing coefficients represent features, detail of the investigated signal, arising at transition from one scale to another and are equal [16]:
 | (4) |
Where
– Detailing wavelet-coefficients 
on level
,
– Scalar product of investigated sequence of data in the form of time series 
and a mother wavelet
on corresponding level of expansion
.
Thus, the main tool for the research of studied processes is processing of the wavelet-coefficients which have been received on different scales. First of all it concerns detailing wavelet-coefficients, which emphasize the characteristics of the time series.
2.2. Basic Properties of Wavelet Coefficients of Self-similar Time Series1. Detailing coefficients of DWS on each level of expansion 
have normal distribution with a zero average
.
2. On each level of expansion 
detailing wavelet-coefficients
, 
are self-similar [16]:
 | (5) |
3. The wavelet-coefficients received as a result of expansion of process with stationary increments are stationary on each scale
.
4. If there are moments of 
order then for the wavelet-coefficients which were received as a result of expansion of process
, the following equality is carried out [16]:
 | (6) |
Where 
– expectation value of the process, which is studied.
The above listed properties of wavelet-expansion can be used for the analysis of various images in practice. For this we look at the various images and the analysis detailing wavelet-coefficients.
2.3. Data for AnalysisFor the analysis, we use a variety of images (these images are in public internet access): Figure 2: Cytological preparation. Mammary gland puncture (image size 1328x996 pixels) and Figure 3: Image for eye fundus (image size 933x937 pixels).
Download as
Download as
The image size determines the length of the time series and the number of levels of wavelet decomposition.
3. Results and Discussion
First of all, we transform the two-dimensional original image (Figure 2 and Figure 3) in the one-dimensional series in accordance with Figure 1. The converted original image (Figure 2 and Figure 3) are shown in Figure 4 and Figure 5, respectively.
Download as
Download as
It is noticeable that the transformed series (Figure 4 and Figure 5) and the original image (Figure 2 and Figure 3) are different, which may allow us to make general conclusions about the possibility of using the wavelet expansion coefficients for image analysis.
In accordance with the size of the original images (Figure 2 and Figure 3) we will make the expansion of one-dimensional series (Figure 4 and Figure 5) into 10 levels with the help of different wavelets (db1, db2 and db4). We will also calculate basic statistical parameters for detailing the coefficients of the wavelet decomposition (expectation value, dispersion, intervals for the parameter that estimates the standard deviation). Thus, for each of source image, we will obtain a specific set of statistical characteristics at each level of decomposition. Table 1 presents the basic statistical characteristics for detailing wavelet coefficients of image in Figure 2, which is represented as a one-dimensional series in Figure 4. Table 2 presents the basic statistical characteristics for detailing wavelet coefficients of image in Figure 3, which is represented as a one-dimensional series in Figure 5. Analysis of Table 1 and Table 2, as well as the detailing wavelet coefficients show that detailing wavelet coefficients of the series, which represent the images at Figure 2 and Figure 3 is the first property of the Wavelet coefficients decomposition.
In Table 1 and Table 2, we see that the expectation value in most cases is zero. Consequently, for detailing the coefficients of wavelet decomposition we have a normal distribution with zero average 
at each level of decomposition. This allows us to use other properties of detailing the coefficients of wavelet decomposition. Given the previously mentioned properties 2 through 4, it is possible to say that each image is characterized by its own set of detailing coefficients at each level of decomposition. Then as individual characteristics of the image can be considered indicators as follows:
Given the properties of 2-4 can say that each image is characterized by its own set of detailing coefficients at each level of decomposition. Then, as individual characteristics of the image the following indicators can be considered:
 | (7) |
 | (8) |
 | (9) |
 | (10) |
Where
– Expected value of detailing wavelet coefficients at the first level,
– Expected value of detailing wavelet coefficients at 
level (
),
– Dispersion of detailing wavelet coefficients at the first level,
– Dispersion of detailing wavelet coefficients at 
level (
),
– Previous and subsequent levels of wavelet decomposition.
In Figure 6 and Figure 7 presented dynamic indicators 
and 
for image in Figure 2 taking into consideration that the decomposition is in different wavelet functions (allocated colors).
Download as
for Figure 2
Download as
for Figure 2
In Figure 8 and Figure 9 presented dynamic indicators 
and 
for image in Figure 3 taking into consideration that the decomposition is in different wavelet functions (allocated colors).
Download as
for Figure 3
Analysis of Figure 6, Figure 7, Figure 8 and Figure 9 shows that changes in dynamic indicators 
and 
for wavelet decomposition at different wavelet functions (db1, db2 and db4) is approximately the same for the same image, and different for different images. At the same time we can say that the indicators 
and 
reflect the various manifestations and features of detailing wavelet coefficients at decomposition at different levels depending on the wavelet function (db1, db2 and db4), which was used to decompose (see Figure 10 through Figure 13).
Download as
for Figure 3
Download as
for Figure 6
Download as
for Figure 7
In Figure 10 and Figure 11 presented dynamic indicators 
and 
for image in Figure 2 taking into consideration that the decomposition is in different wavelet functions (allocated colors).
Download as
for Figure 8
Download as
for Figure 9
In Figure 12 and Figure 13 presented dynamic indicators 
and 
for image in Figure 3 taking into consideration that the decomposition is in different wavelet functions (allocated colors).
Download as
for Figure 2
Download as
for Figure 2
Download as
for Figure 3
Download as
for Figure 3
Analysis of Figure 14, Figure 15, Figure 16 and Figure 17 shows that changes in dynamic indicators 
and 
for wavelet decomposition at different wavelet functions (db1, db2 and db4) is approximately the same for the same image, and different for different images. Therefore, the indicators 
, 
, 
and 
can be used to compare the images with each other, identifying the distinguishing features between them based on the levels of wavelet decomposition, which is used for image analysis (see Figure 18 through Figure 21).
Download as
for Figure 14
Download as
for Figure 15
Download as
for Figure 16
Download as
for Figure 17
4. Conclusions
We have considered the possibility of using the properties of wavelet coefficients decomposition as a comparative analysis for image analysis and processing. For this task we convert two-dimensional images into one-dimensional data sets using multiresolution wavelet analysis, then we find the detailing wavelet coefficients at each level of decomposition for each data set, after that, we build new features that reflect the dynamic change in mean and variance for the detailing wavelet coefficients decomposition at each level of decomposition.
The transformation of the image into new characteristics based on the properties of the wavelet coefficients of decomposition takes into consideration the following:
- Unique representation of the original image as a set of detailing coefficients,
- The presence of the self-similarity of the wavelet coefficients for detailing the various levels of decomposition,
- Conformity of detailing wavelet coefficients to the law of normal distribution at each level of decomposition for all data set.
We also investigated the dynamics of change in mean and variance for the detailing wavelet coefficients decomposition for different wavelet functions. We have also shown that the dynamics are identical to the same images and different for different images, which is allow to build comparison and analysis procedures of how to identify images and compare individual parts. In general, the results have shown the feasibility of using the properties of wavelet coefficients decomposition as a comparative analysis of images.
References
| [1] | Ji, L., & Yi, Z.,“A mixed noise image filtering method using weighted-linking PCNNs”. Neurocomputing, 71(13), 2986-3000, 2008. | ||
 In article | View Article | ||
| [2] | Khan, M. B., Lee, X. Y., Nisar, H., Ng, C. A., Yeap, K. H., & Malik, A. S., “Digital image processing and analysis for activated sludge wastewater treatment”, In Signal and image analysis for biomedical and life sciences , Springer International Publishing, (pp. 227-248), 2015. | ||
| In article | View Article | ||
| [3] | Aksoy, S., Yalniz, I. Z., &Tasdemir, K, “Automatic detection and segmentation of orchards using very high resolution imagery”, IEEE Transactions on geoscience and remote sensing, 50(8), 3117-3131, 2012. | ||
| In article | View Article | ||
| [4] | Pedroso, M., Taylor, J., Tisseyre, B., Charnomordic, B., & Guillaume, S., “A segmentation algorithm for the delineation of agricultural management zones”, Computers and Electronics in Agriculture, 70(1), 199-208, 2010. | ||
| In article | View Article | ||
| [5] | Wang, Y., Wu, G., Chen, G. S., & Chai, T.,”Data mining based noise diagnosis and fuzzy filter design for image processing”. Computers & Electrical Engineering, 40(7), 2038-2049, 2014. | ||
| In article | View Article | ||
| [6] | Javed, U., Riaz, M. M., Ghafoor, A., & Cheema, T. A,“SAR image segmentation based on active contours with fuzzy logic”, IEEE Transactions on Aerospace and Electronic Systems, 52(1), 181-188, 2016. | ||
| In article | View Article | ||
| [7] | Yan, J., Schwartz, L. H., & Zhao, B., “Semiautomatic segmentation of liver metastases on volumetric CT images”, Medical physics, 42(11), 6283-6293, 2015. | ||
| In article | View Article PubMed | ||
| [8] | Gorgel, P., Sertbas, A., &Ucan, O. N., “A wavelet-based mammographic image denoising and enhancement with homomorphic filtering”, Journal of medical systems, 34(6), 993-1002, 2010. | ||
| In article | View Article PubMed | ||
| [9] | Kobylin, O., & Lyashenko, V., “Comparison of standard image edge detection techniques and of method based on wavelet transform”, International Journal of Advanced Research, 2(8), 572-580, 2014. | ||
| In article | |||
| [10] | Lyashenko, V., Matarneh, R., Kobylin, O., & Putyatin, Y., “Contour detection and allocation for cytological images using Wavelet analysis methodology”, International Journal of Advance Research in Computer Science and Management Studies, 4(1), 85-94, 2016. | ||
| In article | |||
| [11] | Kingsbury, N., “Image processing with complex wavelets”, Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 357(1760), 2543-2560, 1999. | ||
| In article | |||
| [12] | Lyashenko, V., Kobylin, O., & Ahmad, M. A., “General Methodology for Implementation of Image Normalization Procedure Using its Wavelet Transform”, International Journal of Science and Research, 3(11), 2870-2877, 2014. | ||
| In article | |||
| [13] | Delbeke, L., & Abry, P., “Stochastic integral representation and properties of the wavelet coefficients of linear fractional stable motion”, Stochastic Processes and their Applications, 86(2), 177-182, 2000. | ||
| In article | View Article | ||
| [14] | Mallat, S., & Hwang, W. L., “Singularity detection and processing with wavelets”, IEEE transactions on information theory, 38(2), 617-643, 1992. | ||
| In article | View Article | ||
| [15] | Flandrin, P., “Wavelet analysis and synthesis of fractional Brownian motion”, IEEE Transactions on information theory, 38(2), 910-917, 1992. | ||
| In article | View Article | ||
| [16] | Lyashenko, V., Deineko, Z., & Ahmad, A., “Properties of wavelet coefficients of self-similar time series”, International Journal of Scientific and Engineering Research, 6(1), 1492-1499, 2015. | ||
| In article | |||
CiteULike
Mendeley
StumbleUpon
Delicious
Facebook
Twitter
LinkedIn
