Image Segmentation by Fuzzy Edge Detection and Region Growing Technique

Amitab Khwairakpam,Ruhul Amin Hazarika,Debdatta Kandar

DOI: https://doi.org/10.1007/978-981-13-7091-5_5

2019-01-01

Abstract:Image segmentation is one of the most important steps in computer vision and commonly used for applications like object detection, object recognition, etc. Some of the frequently used image segmentation techniques include region splitting and merging, k-means clustering, and region growing. Region growing can effectively segment the region of interest. However, due to lack of edge information, region growing process may terminate at the incorrect region boundary. To overcome the termination problem, edge detection and region growing techniques can be combined. In this paper, a new image segmentation technique is proposed. The proposed technique has two basic stages; the first stage detects edges and in the second stage, region growing technique is applied. We have used fuzzy logic to decide edge point in the image. The fuzzy membership function’s parameters are determined by genetic algorithm. In the second stage, seeded region growing technique is applied at the edge detected image, where the seed point and the threshold value for comparing neighboring pixels are computed by the algorithm. The performance of the proposed technique is compared with some of the commonly used existing segmentation techniques based on segmentation accuracy and sensitivity, and it is found that the result of the proposed technique yields better result.

What problem does this paper attempt to address?

The problems that this paper attempts to solve are the inaccurate boundary detection in image segmentation and the problem that the region - growing technique is prone to exceed the boundaries of the target region. Specifically, in the absence of edge information, the traditional region - growing technique may terminate at the wrong position, resulting in inaccurate segmentation results. To solve these problems, the paper proposes a new method that combines fuzzy edge detection and region - growing, aiming to improve the accuracy and sensitivity of image segmentation.

### Specific problems solved by the paper:
1. **Inaccurate edge detection**: Traditional methods may have difficulty in determining whether a point is an edge point when detecting edges due to insufficient pixel information, resulting in low accuracy of edge detection.
2. **Region - growing beyond boundaries**: In the absence of accurate edge information, the region - growing technique may continue to grow outside the target region, resulting in inaccurate segmentation results.
3. **Automated parameter selection**: In traditional methods, the selection of seed points and similarity thresholds usually requires manual intervention, while the algorithm proposed in the paper can automatically select these parameters.

### Solutions:
1. **Fuzzy edge detection**: Use fuzzy logic to detect edge points in an image. Fuzzy logic improves the accuracy of edge detection by defining membership functions to handle the uncertainty of pixels.
2. **Genetic algorithm to optimize parameters**: Use the genetic algorithm to determine the parameters of the fuzzy membership function to optimize the effect of edge detection.
3. **Region - growing**: Apply the region - growing technique on the detected edge image. The seed points and similarity thresholds are automatically selected by the algorithm. The region - growing process will terminate at the edge points, thus avoiding exceeding the target region.

### Experimental results:
The paper verifies the superiority of the proposed method by comparing it with several existing image segmentation techniques (such as K - means clustering, region - splitting - merging, and traditional region - growing). The experimental results show that the proposed method is superior to other methods in terms of accuracy and sensitivity.

### Formula representation:
- **Gaussian membership function**:
\[
\mu(x)=e^{-\frac{(x - x_c)^2}{2\sigma^2}}
\]
where \(x_c\) is the center point and \(\sigma\) is the width.

- **Triangular membership function**:
\[
f(x)=\begin{cases}
0 & \text{if } x\leq a \\
\frac{x - a}{b - a} & \text{if } a\leq x\leq b \\
\frac{c - x}{c - b} & \text{if } b\leq x\leq c \\
0 & \text{if } c\leq x
\end{cases}
\]
where \(a\), \(b\) and \(c\) define the base of the triangle.

- **Accuracy**:
\[
\text{Accuracy}=\frac{TP + TN}{TP + TN + FP + FN}
\]

- **Sensitivity**:
\[
\text{Sensitivity}=\frac{TP}{TP + FN}
\]

Through these methods and techniques, the paper successfully improves the accuracy and reliability of image segmentation.