The week's pick
Random Articles
Reseach Article
Gaussian based Image Segmentation Algorithm
| International Journal of Computer Applications |
| Foundation of Computer Science (FCS), NY, USA |
| Volume 175 - Number 14 |
| Year of Publication: 2020 |
| Authors: Adolf Fenyi, Emmanuel Nkansah, Ebenezer Eghan |
10.5120/ijca2020920615
|
Adolf Fenyi, Emmanuel Nkansah, Ebenezer Eghan . Gaussian based Image Segmentation Algorithm. International Journal of Computer Applications. 175, 14 ( Aug 2020), 10-16. DOI=10.5120/ijca2020920615
Abstract
This study adopted the Gaussian equation to eliminate noisy pixels in a segmented image. The equation was converted into a convolution kernel and the values were normalized to attain a perfect result. In order to preserve every information in the image, the border pixels were padded with zeros before the image was convolved with the designed kernel. A variance was computed from the result obtained from the convolution operation and the pixel that obtained the minimum variance was selected as the threshold. In order to obtain the optimum threshold, the computed threshold was multiplied by the optimization constant which ranges from 0.1 to 1, and the resultant was considered as the final threshold for the segmentation process. The Gaussian algorithm was evaluated with global, Otsu and adaptive algorithms. The performance metrics used were signal to noise ratio (defines the sensitivity of algorithm), mean square error and running time (specifies the execution time of an algorithm). In the first experiment which was made up of 20 noise free images, the adaptive attained the highest rating of 7.117dB. This was followed by the Gaussian algorithm at 5.231dB. The poorest performance was seen in the global at 1.988dB. The second experiment was done with ten noisy images. In the experiment, the Gaussian recorded the highest rating of 3.124dB, while the adaptive scored the poorest at 1.847dB. From the running time, the fastest algorithm was seen in the global at 2.814s. This was followed by the Otsu at 9.814s. There was only a difference of 2.107s between the average execution time of the Otsu and the Gaussian This was due to the noise suppressing mechanism in the Gaussian. The slowest algorithm was the adaptive which recorded a running time of 110.594s The Gaussian algorithm performed better when there was presence of noise in the image. This explains why it recorded sensitivity rate of 3.124dB in the second experiment. However, for noise free images, the adaptive algorithm is the best despite the poor performance in the running time.
References
- Acharya, T., & Ray, A. K. (2005). Image Processing: Principles and Applications (Vol. 15). https://doi.org/10.11 17/1.2348895
- Dass, R., & Devi, S. (2012). Image Segmentation Techniques. International Journal of Electronics & Communication Technology (IJECT), 3(1), 66–70.
- Barghout, L., & Sheynin, J. (2013). Real-world scene perception and perceptual organization: Lessons from Computer Vision. Journal of Vision, 13(9), 700–710.
- ] Shapiro, L. G., & Stockman, C. G. (2001). Computer Vision. New Jersey: Prentice-Hall Publisher.
- Jain, A. (1986). Fundamentals of Digital Image Processing. New Jersey: Prentice-Hall Publisher.
- Davies, E. (1990). Machine Vision: Theory, Algorithms and Practicalities. Cambridge, Massachusetts: Academic Press Publisher.
- Chow, C.K., & Kaneko, T. (1972). Automatic Boundary Detection of the Left Ventricle from Cineangiograms. Computers and Biomedical Research, 4(9), 303–356.
- Rohankar, J. (2013). Survey of the Various Noises and Techniques for Denoising the Colour Image. International Journal of Application or Innovation in Engineering & Managemen, 2(11), 257–289.
- Shapiro, L. G., & Stockman, C. G. (2001). Computer Vision. New Jersey: Prentice-Hall Publisher.
- Sipser, M. (2006). Introduction to the Theory of Computation. Boston MA, USA: Course Technology Publisher.
- Rafael C. Gonzalez, R. E. W. (2001). Digital Image Processing (3rd ed.). Upper Saddle River, New Jersey 07458: Prentice Hall. Retrieved from http://www-cs.ccny.cuny.edu/~zhu/Capstone2007/DIP-Chapter10-Segmentation-notes.pdf
- Bansal, S., & Maini, R. (2013). A Comparative Analysis of Iterative and Ostu ’ s Thresholding Techniques. International Journal of Computer Applications, 66(12), 45–47.
- Rosenfeld, Y. N. and A. (1979). Some experiments on variable thresholding. Pattern Recognition, 11(3):191–204.
- Rafael C. Gonzalez, R. E. W. (2001). Digital Image Processing (3rd ed.). Upper Saddle River, New Jersey 07458: Prentice Hall. Retrieved from http://www-cs.ccny.cuny.edu/~zhu/Capstone2007/DIP-Chapter10-Segmentation-notes.pdf
- Sunil L. Bangare, Amruta Dubal, Pallavi S. Bangare, D. S. T. P. (2015). Reviewing otsu ’ s method for image thresholding, (August 2016).
- Vala, M. H. J., & Baxi, A. (2013). A review on Otsu image segmentation algorithm. International Journal of Advanced Research in Computer Engineering and Technology, 2(2), 387–389.
- Martin, D., Fowlkes, C., Tal, D., & Malik, J. (2001). A Database of Human Segmented Images and its Application to Evaluate Segmentation Algorithms and Measuring Ecological Statistics. Proceedings of 8th International Conference on Computer Vision (pp. 416--423).
Index Terms
Keywords