Image segmentation is an important image technique well known by its utility and complexity. To extract the useful information from images or groups of images, an inevitable step is to separate the objects from the background. Segmentation is just the right process and technique required for this task. Image segmentation is often described as the process that subdivides an image into its constituent parts and extracts those parts of interest (objects). It is one of the most critical tasks in automatic image analysis, which is at the middle layer of image engineering. Image engineering (which is composed of three layers from bottom to top: (1) image processing, (2) image analysis, and (3) image understanding) is a new discipline and a general framework for all image techniques (Zhang, forthcoming). The history of segmentation of digital images using computers can be traced back to 40 years ago. In 1965, an operator for detecting the edges between different parts of an image, Roberts operator (also called Roberts edge detector), was introduced and used for partition of image components (Roberts, 1965). Since then, the field of image segmentation has evolved very quickly and has undergone great change (Zhang, 2001a). In this article, after an introduction and explanation of the formal definition of image segmentation as well as three levels of research on image segmentation, the statistics for the number of developed algorithms in these years are provided; the scheme for classifying different segmentation algorithms is discussed; and a summary of existing survey papers for image segmentation is presented. All these discussions provide a general picture of research and development of image segmentation in the last 40 years.
Formal Definition of Image Segmentation
A formal definition of image segmentation, supposing the whole image is represented by R and Ri,i = 1, 2, …, n are disjoint nonempty regions of R, consists of the following conditions (Fu & Mui, 1981):; (1) For all i and j, i ≠ j, there exits = ∅; (2) For i = 1, 2, …, n, it must have P(Ri) = TRUE; (3) For all i ≠ j, there exits P () = FALSE; (4) where P(Ri) is a uniformity predicate for all elements in set Ri and ∅ represents an empty set.
The following condition is also important for segmentation and is often included in the conditions for the formal definition (Zhang 2001a):
For all i
= 1, 2, …, n
is a connected component. (5)
In the aforementioned conditions, each of them has particular meanings. The condition (1) points out that the union of segmented regions could include all pixels in an image. The condition (2) points out that the different segmented regions could not overlap each other. The condition (3) points out that the pixels in the same regions should have some similar properties. The condition (4) points out that the pixel belonging to different regions should have some different properties. The condition (5) points out that the pixels in the same region resulted from segmentation are connected.
Key Terms in this Chapter
Active Contour Model: Active contour model is a sequential technique for image segmentation. Given an approximation of the boundary of an object in an image, an active contour model can be used to find the actual boundary by deforming the initial boundary to lock onto features of interest within this image.
Region Growing: Region growing is a region-based sequential technique for image segmentation by assembling pixels into larger regions based on predefined seed pixels, growing criteria, and stop conditions.
Watersheds: Watershed technique is inspired from the topographic interpretation of image segmentation by watersheds embodies many concepts of edge detection, thresholding and region processing techniques, and often produces stable and continuous results.
Image Segmentation: A process consists of subdividing an image into its constituent parts and extracting these parts of interest (objects) from the image.
Edge Detection: Edge detection is the most common approach for detecting discontinuities in images, and is the fundamental step in edge-based parallel process for segmentation. An edge is a local concept. To form a complete boundary of an object, edge detection should be followed by edge linking or connection.
Thresholding: Thresholding techniques are the most popularly used segmentation techniques. A set of suitable thresholds need to be first determined, and then the image can be segmented by comparing the pixel properties with these thresholds.
Image Engineering: Image engineering is an integrated discipline/subject comprising the study of all the different branches of image and video techniques. It mainly consists of three levels: image processing, image analysis, and image understanding.
Graph Search: Graph search is a particular type of segmentation technique which combines edge detection and linking together. It represents edge segments in the form of a graph and searches the graph for low-cost paths that correspond to significant edges or boundaries of objects.
Clustering: Clustering is also called unsupervised learning and is a powerful technique for pattern classification. It is a process to group, based on some defined criteria, two or more terms together to form a large collection In the context of image segmentation, it is often considered as the multi-dimensional extension of the thresholding technique.
Gradient Operator: Gradient operator is the first type of operator used for edge detection. The gradient of an image is a vector consisting of the first order derivatives (including the magnitude and direction) of an image.