Journal of Global Research in Computer Sciences
MenuISSN: 2229-371X
ISSN: 2229-371X
Debabrata Samanta*1, Goutam Sanyal2
|
| Corresponding Author: Debabrata Samanta, E-mail: debabrata.samanta369@gmail.com |
| Related article at Pubmed, Scholar Google |
Visit for more related articles at Journal of Global Research in Computer Sciences
The statistical parameters include high order image statistics like skewness which describe the shape and symmetry of the image. SAR images data are the result of a coherent imaging system that produces the impair noise fact. Image segmentation is the process of unscrambling or grouping an image into different parts. The good feat of recognition algorithms based on the quality of segmented image. The main problem in SAR image application is correct segmentation. In this paper, we consider the problem of SAR image segmentation by a statistical parameter.
INTRODUCTION |
| SAR images come across gradually more wide applications, as SAR sensor does not based on weather condition and can infiltrate clouds. SAR image Segmentation is a main examples of low-level operators that providing the basic information for classification of different region. Expected values and variances are important concepts in statistics. They are used to describe distributions, to evaluate the performance of estimators, to obtain test statistics in hypothesis testing, and many other applications. The motivation of this work is to develop a novel segmentation algorithm, which can be used to segment the SAR images and recover the overall accuracy. There are different methods of segmenting an image based on the intensity value, including a simple thresholding, quantization of the image, etc. The important factor in this method of segmentation is choosing a threshold for classification of a region to a particular segment. Deciding on a generic threshold that would work for all images is essentially impossible and hence segmenting an image based on just a pre-defined threshold is not a practical technique. Segmentation algorithms for Synthetic Aperture Radar images have been suggested in the field of categorization. C. H. Fosgate, W. W. Irving and W.C.Karl proposed statistical differences in SAR imagery of different terrain types. Lionel Bombrun used heterogeneous clutter models to describe the SAR data. A.A.G., C.L.M. used multi-scale PolSAR data filtering technique based on BPT for SAR image segmentation E.V.D.Lucca and A.C.Frery proposed MUM and RWSEG general means for better segmentation. |
PROPOSED METHOD |
| A. Segmentation based on gray level intensity |
| Generally the popular method of segmenting an image is based on the gray scale intensity values in the image. Segmenting an image into different regions is done by clustering each gray level intensity value tightly to a mean value or centroid. This is an iterative method, which needs only the number of segments to be hard-coded and adaptively finds the threshold to classify regions within the image. As an initial approximation, the number of segments is predefined as N, and as many different, equally spaced mean values are obtained based on the maximum and minimum gray level values in the image, as given in the following equation |
 |
| max is the maximum gray level intensity value, and min is the minimum gray level intensity value. Each of the pixel values in the image is now assigned to the bin of the corresponding class based on their distance from the mean value of each class. As a measure to identify the distance, the linear distance is calculated as |
 |
| Where Ix,y, is the pixel intensity at the location (x,y). After all the pixels have been placed into a class, the mean value of all the pixels in a given class is evaluated and used as the new reference mean for assigning pixels in the next iteration. This process is repeated until the mean values converge or a given number of iterations have been completed. Finally, the image is classified into N regions based on the mean values. The value of N is chosen based on the number of features that require extraction and their corresponding gray level intensities. Features of interest to us in the images are houses, roads, roofs, sidewalks, driveways and trees. |
| B. Mean |
| The statistical mean gives us the average value of all the pixels for a particular feature. Mathematically, the statistical mean of a set of variables is expressed as |
 |
| The summation is carried out over all the gray level values in the set. It is important to notice that, even though the mean gives a good estimate of the characteristics of a feature, it might not give a good representation of the damage or change in the region. |
| C. Variance |
| Variance is the second moment of a matrix and is a measure of gray-level contrast that can be used to establish descriptors of relative smoothness. Mathematically, it is defined as the mean of squares of differences between respective samples (graylevel intensities) and their mean and is expressed as |
 |
| Where δ2 the variance or square of the standard deviation, and n is the number of samples in the region. An estimate of the difference in the variance values of the region between the predicator and post-disaster images is an indication of the level of change. |
| D. Entropy |
| Entropy is a measure of disorder, or more precisely unpredictability. The probability of a SAR images intensity occurring at particular pixel in ’k’, where ’k’ is the set of all pixels in a SAR image, is defined as p{n}log(p{n}) . The sum of all of these probability makes the Entropy of ’P’, so, |
 |
| Where p{n} is the probability mass function of particular pixel in ’k’. |
| E. Median |
| The median of a variable is defined as the middlemost value when its values are arranged in ascending or descending order of magnitude. In other hand, the median divides the whole set of binary image values in two parts such that half of the observations are less than or equal to it and half are more than equal to it. So median is, |
 |
| and with class interval, respectively. |
| F. Mode |
| The mode of a variable is that values of the variable which has the height frequency or frequency density. A binary image can be fitted with a smooth frequency curve, and then the mode is the abscissa of the height point on the curve. So the mode is , |
 |
| Where, C is the common with of the class. |
| G. Skewness |
| Skewness, or the third standardized moment, is a measure of the asymmetry of the probability distribution of a real valued random variable. |
| If skewness is negative, the data are spread out more to the left of the mean than to the right. If skewness is positive, the data are spread out more to the right. The skewness of the normal distribution (or any perfectly symmetric distribution) is zero. |
| First order Skewness: When the frequency distribution is not symmetrical, it depends on first order Skewness. So, first order skewness given below: |
 |
| Skewness characterizes the degree of asymmetry of a distribution around its mean. The skewness of a distribution is a non-dimensional quantity calculated as |
 |
| Where δ is the standard deviation or the square root of the variance. A positive value indicates a deviation of the distribution more towards the brighter side and a negative value towards the darker side. A distribution is symmetrical if it shows two identical mirror images when it is split down the middle as the normal distribution when a distribution is not symmetrical, it has elongated tail at one side (left or right) and there are more data in this side tail that would be expected in a normal distribution. When it happens at the left, the distribution is negatively skewed or skewed to the left, and the in same way, when it happens at the right the distribution is positively skewed or skewed to the right. |
 |
| H. Kurtosis: |
| The fourth standardized central moment, or kurtosis, is generally defined as a measure reflecting the degree to which a distribution is peaked. In particular, kurtosis provides information regarding the height of the distribution relative to the value of its standard deviation. The kurtosis of the normal distribution is 3. Distributions that are more outlier-prone than the normal distribution have kurtosis greater than 3; distributions that are less outlier-prone have kurtosis less than 3. The kurtosis of a distribution is defined as |
 |
| Where μ is the mean of x, is the standard deviation of y, and E(t) represents the expected value of the quantity t. The kurtosis value (K) can be classified in three general categories according to its derivation from a normal distribution: mesokurtic (normal distribution, K=0), leptokurtic (high degree of peakedness, K>0) and platykurtic (low degree of peakedness, K<0). Figure 2 illustrates these three distribution types. The kurtogram will provide an indication of, first, the location of eventual subsurface defects, and second, their thermal diffusivity. |
 |
K-MEANS CLUSTERING ALGORITHM |
| K-means clustering algorithm is a method of cluster analysis which aims to partition n observations into k clusters in which each observation belongs to the cluster with the nearest mean. K-means clustering Algorithm, known as also C-means clustering, has been applied to variety of areas, including image and speech data compression. The objective function is defined as |
 |
| data point Ei (j) and the cluster centre Vj is an indicator of the distance of the n data points from their respective cluster centers. |
PROPOSED ALGORITHM |
| Input: SAR Images of variable size. |
| Output: Segmented region of SAR image. |
| 1) Start. |
| 2) Taken a SAR images. |
| 3) Consider a 3X3 window. |
| 4) Calculate the mean, variance SAR Images. |
| 5) Store the color feature in the standardized moment i.e. skewness and fourth standardized central moment i.e. kurtosis. |
| 6) Segmentation is obtained using K-means clustering Algorithm. |
| 7) Stop. |
EXPERIMENTAL RESULT |
| Here SAR images are used to test proposed methodology. At first, the mean, variance was calculated for original image. Images were tested for analyzing window size of 3x3 and 5x5 to find the third standardized moment, fourth standardized central moment. Finally K-means clustering Algorithm was used on Skewness image for segmentation. Now we calculate the features table given below. |
 |
 |
CONCLUSION AND DISCUSSION |
| In this paper, we proposed a novel statistical parameter based on skewness and kurtosis for segmentation of SAR image. This technique based on considering a 3X3 window and calculates the mean, variance, mode, median, entropy of that SAR Images. Next, Segmentation is obtained using K-means clustering Algorithm. It was evidenced that this segmentation procedure is a straightforward extension of the filtering algorithm based on standardized moment i.e. skewness. For this reason, our algorithm did not make mistakes; that is, a segmented image very different to get the originality of the SAR images. This may be extended to the color image segmentation. The results from this preliminary study indicated that the proposed strategy was effective. |
References |
|