DATA STRUCTURES FOR IMAGE ANALYSIS
Data and an algorithm are the two essentials of any program. Data organization often considerably
affects the simplicity of the selection and the implementation of an algorithm, and the choice of data
structures is therefore a fundamental question when writing a program.
1.8: LEVELS OF IMAGE DATA REPRESENTATION
The aim of computer visual perception is to find a relation between an input image and models of
the real world. During the transition from the raw input image to the model, image information
becomes denser and semantic knowledge about the interpretation of image data is used more.
Several levels of visual information representation are defined on the way between the input image
and the model.
The representations can be stratified in four levels [Ballard and Brown, 1982]. These four
representational levels are ordered from signals at a low level of abstraction to the description that a
human can perceive. The information flow between the levels may be bi-directional.
1) The lowest representational level - iconic images consist of images containing original data:
integer matrices with data about pixel brightness.
2) The second level is segmented images. Parts of the image are joined into groups that probably
belong to the same objects.
3) The third level is geometric representations holding knowledge about 2D and 3D shapes.
4) The fourth representational level is relational models. They give us the ability to treat data more
efficiently and at a higher level of abstraction.
1.9/TRADITIONAL IMAGE DATA STRUCTURES
Traditional image data structures such as matrices, chains, graphs, lists of object properties, and
relational databases are important.
1.9.1. Matrices
A matrix is the most common data structure for low-level representation of an image. Elements of
the matrix are integer numbers corresponding to brightness, or to another property of the
corresponding pixel of the sampling grid.
Chains
Chains are used for the description of object borders in computer vision. One element of the chain is
a basic symbol: this approach permits the application of formal language theory for computer vision
tasks.
Chain codes are often used for the description of object borders or other one-pixel-wide lines in
images. The border is defined by the coordinates of its reference pixel and the sequence of symbols
corresponding to the line.