Unlock instant, AI-driven research and patent intelligence for your innovation.

Fast robust fuzzy C-means image segmentation method combining neighborhood information

A technology of neighborhood information and average image, which is applied in image analysis, image data processing, instruments, etc., and can solve problems such as blurring, pixel information loss, and irrationality

Inactive Publication Date: 2012-10-24
NANJING NORMAL UNIVERSITY
View PDF5 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

When there are pixels seriously polluted by noise in the image, the value of α must be set large enough to resist the influence of these noise points, but for pixels that have not been polluted by noise or are slightly polluted in the same image, an excessively large α inevitably causes the information of these pixels to be lost or blurred
Therefore, in the segmentation process, it is unreasonable to set the same size of α for all neighborhood windows

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Fast robust fuzzy C-means image segmentation method combining neighborhood information
  • Fast robust fuzzy C-means image segmentation method combining neighborhood information
  • Fast robust fuzzy C-means image segmentation method combining neighborhood information

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0067] figure 2 (a) is a 128×128 artificial image containing two clusters with gray values ​​0 and 90. In order to test the robustness of the algorithm, Gaussian noise, salt and pepper noise and mixed noise are respectively added to the image, where the mixed noise is a combination of Gaussian noise N(0,100) and SαS noise, figure 2 (b) is a noisy image. figure 2 (c)-(f) show the segmentation results of FCM_S1, FCM_S2, EnFCM and FGFCM algorithms in the case of mixed noise (α=0.7 in SαS). The segmentation results of FCM_S1, _S2 and EnFCM are all affected by noise points to varying degrees, which shows that these three algorithms lack the ability to resist mixed noise. In contrast, the FGFCM method is hardly affected by noise.

Embodiment 2

[0069] image 3 (a) is the noise-free real image ‘Eight’ with a size of 308×242. image 3 (b) is the image polluted by mixed noise. image 3 (c)-(f) are the comparison of segmentation results of FCM_S1, FCM_S2, EnFCM and FGFCM algorithms on noisy images. Visually, FCM_S1, FCM_S2, and EnFCM algorithms are affected by noise to varying degrees, while FGFCM can basically eliminate the influence of mixed noise.

Embodiment 3

[0071] Comparing the robustness of segmentation results to noise and the preservation of details on the noisy medical image 'Brain MR'. Figure 4 (a) is the original 'Brain MR' image, and (b) is the image after adding mixed noise. Figure 4 (c)–(f) are the segmentation results of FCM_S1, FCM_S2, EnFCM and FGFCM algorithms using a neighborhood window of 3×3. It can be observed from the segmentation results that FCM_S1 is affected by noise, and almost none of the other algorithms are affected by noise. However, the segmentation results of FCM_S2 and EnFCM have varying degrees of blur. In contrast, FGFCM can not only resist the influence of noise points, but also retain as much detail information of the image as possible, especially in the part of the ellipse mark.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to a fast robust fuzzy C-means image segmentation method combining neighborhood information. The method mainly comprises the following two steps: first, reshaping image grey scale according to the local correlation of an image; and then performing a rapid fuzzy C-means segmentation algorithm on the grey scale-reshaped image. The fast and robust fuzzy C-means image segmentation method combining neighborhood information provided by the invention has the following two predominant features: (1) a similarity measurement is designed through grey scale correlation and spatial correlation between image pixels, and the double purposes of removing picture noise and retaining image details are achieved by using the measurement; and (2) pixel-based segmentation is converted into gray scale-based segmentation by using the distribution characteristics of grey scale values, and the corresponding time complexity is reduced from 0(NcI1) to 0(QcI2), where c is the number of clusters, I1 and I2 are respectively the numbers of iterative steps of pixel segmentation and grey segmentation, and as the number of grey levels Q is far smaller than the number of pixels N, the time complexity in the segmentation stage is greatly reduced by using the method provided by the invention.

Description

technical field [0001] The invention belongs to the field of image segmentation, in particular to an image segmentation method based on fuzzy C-means. Background technique [0002] Since image segmentation can be regarded as a clustering process of image pixels, clustering methods such as Fuzzy C-Means (FCM) can be directly applied to the field of image segmentation. However, when the image is blurred and noisy, the direct clustering of pixels often cannot obtain satisfactory results. In order to improve the segmentation performance of FCM, many researchers introduce the spatial position relationship of images into the original FCM. Tolias and Panas used a Sugeno-type rule system to impose spatial continuity constraints on the images to be segmented. Pham introduces the spatial constraints of fuzzy membership in the objective function of FCM, so that the fuzzy membership of pixels has spatial smoothness. In recent years, Chen et al. introduced membership constraints and s...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/00
Inventor 蔡维玲杨明
Owner NANJING NORMAL UNIVERSITY