Image Data Structures:Introduction and What is Image Data?

Introduction

Image has been an integral part of our communication. Visual information aids us in understanding our surroundings better. Image processing, the science of manipulating digital images, is one of the methods used for digitally interpreting images. Image processing generally comprises three main steps:

1. Image acquisition: Obtaining the image by scanning it or by capturing it through some sensors.

2. Image manipulation/analysis: Enhancing and/or compressing the image for its transfer or storage.

3. Display of the processed image.

Image processing has been classified into two levels: low-level image processing and high- level image processing. Low-level image processing needs little information about the con- tent or the semantics of the image. It is mainly concerned with retrieving low-level descriptions of the image and processing them. Low-level data include matrix representation of the actual image. Image calibration and image enhancement are examples of low-level image processing.

High-level image processing is basically concerned with segmenting an image into objects or regions. It makes decisions according to the information contained in the image. High-level data are represented in symbolic form. The data include features of the im- age such as object size, shape and its relation with other objects in the image. These image-processing techniques depend significantly on the underlying image data structure.

FIGURE 57.1: Example of an image.

Efficient data structures for region representation are important for use in manipulating pictorial information.

Many techniques have been developed for representing pictorial information in image processing [20]. These techniques include data structures such as quadtrees, linear quadtrees, Forest of Quadtrees, and Translation Invariant Data structure (TID). We will discuss these data structures in the following sections. Research on quadtrees has produced several interesting results in different areas of image processing [2] [6] [7] [11] [12] [14]. In 1981, Jones and Iyengar [8] proposed methods of refining quadtrees. A good tracing of the history of the evolution of quadtrees is provided by Klinger and Dyer [13]. this handbook.

What is Image Data?

An image is a visual reproduction of an object using an optical or electronic device. Image data include pictures taken by satellites, scanned images, aerial photographs and other digital photographs. In the computer, image is represented as a data file that consists of a rectangular array of picture elements called pixels. This rectangular array of pixels is also called a raster image. The pixels are the smallest programmable visual unit. The size of the pixel depends on the resolution of the monitor. The resolution can be defined as the number of pixels present on the horizontal axis and vertical axis of a display monitor. When the resolution is set to maximum the pixel size is equal to a dot on the monitor. The pixel size increases with the decrease of the resolution. Figure 57.1 shows an example of an image.

In a monochrome image each pixel has its own brightness value ranging from 0 (black) to 255 (white). For a color image each pixel has a brightness value and a RGB color value. RGB is an additive color state that has separate values for red, green, and blue. Hence each pixel has independent values (0 to 255) for red, green, and blue colors. If the values for red, green, and blue components of the pixel are the same then the resulting pixel color is gray. Different shades of gray pixels constitute a gray-scale image. If pixels of an image have only two states, black or white, then the image is called a binary image.

Image data can be classified into raster graphics and vector graphics. Raster graphics, also known as bitmap graphics, represents an image using x and y coordinates (for 2d- images) of display space; this grid of x and y coordinates is called the raster.

Figure 57.2 shows how a circle will appear in raster graphics. In raster graphics an image is divided up

FIGURE 57.2: A circle in raster graphics.

in raster. All raster dots that are more than half full are displayed as black dots and the rest as white dots. This results in step like edges as shown in the figure. The appearance of jagged edges can be minimized by reducing the size of the raster dots. Reducing the size of the dots will increase the number of pixels needed but increases the size of the storage space.

Vector graphics uses mathematical formulas to define the image in a two-dimensional or three-dimensional space. In a vector graphics file an image is represented as a sequence of vector statements instead of bits, as in bitmap files. Thus it needs just minimal amount of information to draw shapes and therefore the files are smaller in size compared to raster graphics files.

Vector graphics does not consist of black and white pixels but is made of objects like line, circle, and rectangle. The other advantage of vector graphics is that it is flexible, so it can be resized without loss of information. Vector graphics is typically used for simple shapes. CorelDraw images, PDF, and PostScript are all in vector image format. The main drawbacks of vector graphics is that it needs longer computational time and has very limited choice of shapes compared to raster graphics.

Raster file, on the other hand, is usually larger than vector graphics file and is difficult to modify without loss of information. Scaling a raster file will result in loss of quality whereas vector graphics can be scaled to the quality of the device on which it is rendered. Raster graphics is used for representing continuous numeric values unlike vector graphics, which is used for discrete features. Examples of raster graphic formats are GIF, JPEG, and PNG.