[attachment=14489]
SEGMENTATION TECHNIQUES FOR UNDERWATER IMAGES
The quality of underwater images is directly affected by various mediums. This emphasizes the necessity of application of Image Segmentation algorithms.
With the help of quantitative measures, the results of various segmentation algorithms are compared.
Introduction :
Image segmentation:
A technique that decomposes an image into meaningful parts or objects.
The result of this technique is a set of segments that collectively cover the entire image, or a set of contours extracted from the image.
2.Canny Edge Operator :
Canny operator finds edges by looking for local maxima of the gradient of Image. The gradient is calculated using the derivative of a Gaussian filter.
This method uses two thresholds to detect both strong as well as weak edges and include weak edges in the output only if it is connected to strong edge. This method is therefore is less likely to be fooled by noise than the others, and more likely to detect true weak edges.
Comparison of Sobel and Canny Operators
The algorithmic steps for canny edge detection technique are as follows:
Edge Detection final results :
Adaptive Thresholding :
Adaptive thresholding is a technique which uses different thresholds for different regions in an image.
The output of the Thresholding operation is a binary image whose one part indicates the object and the other the background.
Divide the image into sub-images.
Assume that the illumination in each sub-images is constant.
Use a different threshold for each sub-image.
The algorithmic steps for Adaptive Thresholding Technique :
1. Select an initial estimate for T (typically the average grey level in the image)
2. Segment the image using T to produce two groups of pixels: G1 consisting of pixels with grey levels >T and G2 consisting pixels with grey levels ≤ T
3. Compute the average grey levels of pixels in G1 to give μ1 and G2 to give μ2
The algorithmic steps for Adaptive Thresholding Technique :
Adaptive Thresholding Final Results :
Fuzzy C Means Clustering Method :
Clustering ????
Method by which large sets of data is grouped into clusters of smaller sets of similar data..
There are two types of clustering approaches namely
Hard Clustering
Here data is divided into a number of distinct clusters, where each data element belongs to exactly one cluster.
Soft/Fuzzy Clustering
Here data elements can belong to more than one cluster with a degree of some membership values.
Continue…..
Fuzzy C-Means clustering algorithm (FCM), put forward by Bezdek in 1973, is the improved hard c-means clustering method.
Fuzzy c-means (FCM) is a method of clustering which allows one piece of data to belong to two or more clusters.
By iteratively updating the cluster centers and the membership grades for each data point, FCM iteratively moves the cluster centers to the "right“ location within a data set
The Algorithm steps for Fuzzy C Means Clustering Method :
1. Initialize U=[uij] matrix, U(0)
2. At k-step: calculate the centers vectors
C(k)=[cj] with U(k)
3. Update U(k) , U(k+1)
4. If || U(k+1) - U(k)||< threshold then
STOP; otherwise return to step 2.
Fuzzy C Means Clustering Method Final Result :
Final Results of all the algorithm :
Non linear objective assessment:
Probability function:
The probability function of gray level image is estimated from its histogram, which is formulated as,
P(x) = h(x) / ∑h(x)
Where p(x) is the probability distribution function and
h(x) is the histogram function.
It can be with subjective assessment that the fuzzy c means clustering distribution is close to the original image distribution.
To estimate the quality of the reconstructed images, following objective assessment parameters are used.
Energy:
The gray level energy indicates how the gray levels are distributed. It is formulated as
E(x) = ∑ p(x)
where E(x) represents the gray level energy and p(i) refers to the probability distribution functions.
The larger energy value corresponds to the lower number of gray levels, which means simple. The smaller energy corresponds to the higher number of gray levels, which means complex.
Comparison of different algorithms for underwater Images based on Energy
Discrete Entropy :
The discrete entropy is the measure of image information content, which is interpreted as the average uncertainty of information source.
It is calculated as the summation of the products of the probability of outcome multiplied by the log of the inverse of the outcome probability.
It is formulated as
H(X) =
H(X) =
Comparison of different algorithms for underwater Images based on DiscreteEntropy
Mutual information:
Image normalization using mutual information refers to eliminating image variations (such as noise, illumination, or occlusion) that are irrelevant to object identity.
The goal is to obtain a standard image with no artefacts arising from the specific conditions in which a particular image was taken.
The mutual information of two discrete random variables X and Y can be defined as:
Comparison of different algorithms for underwater Images based on Mutual Information
Normalized mutual information:
The normalized mutual information is a well defined measure covering contents from both discrete entropies and mutual information
It is formulated as
NMI
Comparison of different algorithms for underwater Images based onNormalized Mutual Information
Final Results of all the algorithm :
Final Results of all the algorithm :
Comparison of QM for underwater Titanic Ship Image
Comparison of QM for underwater wreckage Image
Comparison of QM for underwater aquarious skerry Image
Conclusion:
Adaptive thresholding by otsu’s method also had a better segmentation outputs but the depth, shape and curvatures in an image was visible clearly in fuzzy c-means segmentation method.
Hence after evaluation of various non-linear objective assessments, fuzzy c means is the best suited segmentation technique.
APPLICATIONS OF UNDERWATER IMAGE SEGMENTATION
Feature extraction has numerous applications on telecommunication, weather forecasting, environment exploration and medical diagnosis.
Image segmentation can be used for image recognition, estimation with in motion or stereo systems and image compression.
Underwater image taken by the submarines in motion can be effectively segmented, that helps in detecting the threats or enemy’s in water which is not visible in the image due to rapid light fluctuations and current flow of water.
Segmentation of underwater images also helps detecting submerged objects, corals, and rare specimens under water.
SCOPE OF UNDERWATER IMAGE SEGMENTATION
In future ,
Fuzzy C means clustering method with Thresholding gives desirable results when compare to Fuzzy clustering method.
LIMITATIONS OF UNDERWATER IMAGE SEGMENTATION
Fuzzy segmentation is its heavy burden on computing time.
It is not benefit for real time process.