Self-adaptive fuzzy C-means clustering noise image segmentation method and system
An adaptive fuzzy and mean value clustering technology, applied in image analysis, image enhancement, image data processing, etc., can solve problems such as difficult parameter selection, inflexible methods, and impact on segmentation effects, so as to reduce the number of iterations and achieve accuracy and flexibility, the effect of improving segmentation efficiency
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
example 1
[0075] like figure 2 As shown, the image segmentation steps are as follows:
[0076] Step 1: Input the noise image, the noise image such as image 3 As shown in (a), the noise image is a fingerprint image (798×958 pixels in size) with mixed noise added. First, the image is normalized to the range of [0,1], and then the standard deviation of the noise component is 0.15. Gaussian white noise, salt and pepper noise with a density of 0.15, and mean noise with a standard deviation of 0.15. The experimental parameters are set as follows: the total number of clusters K = 2, the membership weight index m = 2, the error ε = 0.001, a very small number eps=0.000001, geometric distribution σ d =3.0 and the luminosity distribution σ r =6.0;
[0077] Step 2: According to the geometric distribution σ d and the luminosity distribution σ r , apply a fast bilateral filter to obtain a filtered image, where the fast bilateral filter is:
[0078]
[0079]
[0080] Among them, δ(ζ-i q ...
example 2
[0106] Segment the street noise image using the same steps as in Example 1. The street noise image is a street image (with a size of 506 × 507 pixels) with mixed noise added. First, the image is normalized to the range of [0, 1], and then the Gaussian white noise with a standard deviation of 0.15 of the noise component is added, and the density is Salt and pepper noise of 0.15 and mean noise with standard deviation of 0.15, the experimental parameters are as follows: the total number of clusters K = 2, the membership weight index m = 2, the error ε = 0.001, the small number eps = 0.000001, the geometric distribution σ d =6.0 and the luminosity distribution σ r = 12.0. like Figure 5 (a).
[0107] Figure 5 (b) is the manually segmented reference image, Figure 5 (c) is the image segmentation result obtained by the traditional FCM clustering method, Figure 5 (d) is the image segmentation result obtained by adopting the method provided by the present invention, from Fig...
example 3
[0113] Using the same procedure as in Example 1, for Figure 7 (a) Segmentation is performed. Figure 7 (a) is an artificial grayscale image (size is 700×700 pixels), first normalize the image to [0,1] range, and then add Gaussian white noise with a standard deviation of 0.15 and a density of 0.15 The salt and pepper noise and the mean noise with a standard deviation of 0.15, the experimental parameters are as follows: the total number of clusters K = 3, the weight index of membership degree m = 2, the error ε = 0.001, a very small number eps = 0.000001, the geometric distribution σ d = 3.5 and the photometric distribution σ r = 3.0. The object is segmented into three types of regions, namely background, hub and rim.
[0114] from Figure 7 (c) It can be seen that the traditional FCM clustering segmentation results are not much different from the original noise image visually due to noise interference, and are different from the original noise image. Figure 7 (b) For com...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More