In the previous article, we have discussed about different types of vector data model and data structures. In this article we will discuss about various types of raster data model, its advantages and disadvantages. We will also learn about various data compression techniques used for raster data structure.
Table of Contents
What is Raster Data Model?
Raster data model is conceptual model used in GIS for representing real world entities or phenomena.
Raster data model makes use of regular array or matrix of cells (of uniform size and shape) for representing data. The cells contain values or attributes for showing the variation of features.
The digital photographs we see in our everyday life, satellite imagery, aerial photographs etc., are some of the good examples of raster data model.
Where do we use Raster Data Model?
Raster data model is widely used for representing continuous features such as elevation, rainfall, temperature etc.
Advantages and Disadvantages of Raster Data Model:
Following are various advantages and disadvantages of raster data structure.
It is very simple data structure.
Continuous features are best represented using raster.
Overlay analysis is easy to perform with raster model.
Topology is not present and has to be represented explicitly.
Huge storage requirement even for storing simple data.
For storing multiple attributes at a given cell, multi band data set is required.
Small features or details are often not observed in the data set depending on spatial resolution (pixel size).
Different Types of Raster Models (or Raster Representations):
Raster data is usually represented in the form of regular array of grids covering the study region (or area of interest). However, it is possible to represent the data present in the study region in different ways as shown below.
Entities are real world objects (or phenomena such as rainfall, humidity etc.). In entity model, the pixels with same entity are shown with same gray level or color.
Cell by Value Model or Pixel Value Model:
In pixel value model, each entity in the study area is assigned with a number. The pixels with same entities are stored with same numbers or values instead of gray levels.
File Model or raster file structure:
Instead of constructing a image with pixels, the details of the study area are stored in a file structure. However, the study area is divided into a matrix of uniform grids, similar to the above models.
Number of rows and columns forming the matrix, different values (as in pixel value model) for each cell are stored in a simple file.
Raster Data Compression Models (Raster Data Encoding):
One of the main disadvantage of raster data model is data storage. Huge space is required for storing data depending on the resolution of raster. For finer resolution and multi band data sets, the storage requirement grows exponentially.
This problem can be solved by using various compression models developed for raster data structure. All these models basically use a file structure for storing the data. However, for optimizing the storage different models use different types of data encoding methods.
Following are the most important and widely used raster data encoding methods.
Run Length Encoding:
In this methods, each row in the image is checked for a group of similar pixels. Instead of storing all values in a group, a single value is used for entire group.
Following procedure explains run length encoding.
Details of raster are encoded using file structure.
First line in the file represents size of the image (no of rows and columns), and no of different features in the study area.
From second row on wards, the details of objects are stored using a pair of numbers for each group of similar values.
The first number in the pair indicates the object id, and second number indicates no of pixels in the row representing the object.
File structure is used for storing the image data
Here data reduction happens in 2 dimensions (along row and column) at a time
The image matrix will have numbering for rows and columns to identify pixels.
Here Square shaped block (same no of rows and columns) of pixels having same data re identified and assigned with a value.
Image size and no of different features in the image are stored in the first row of file.
After that, block size (4×4, 3×3, 2×2, 1×1 etc.,), no of blocks and the starting pixel position of each block are stored in the subsequent lines.
File structure is used for storing the data of image.
Here, the boundary of a region (group of same pixels) is stored using a chain of pixels.
The coordinates of the chain are recorded using several pairs of values, where first number in the pair indicates the direction of movement of chain and second number indicates number of pixels in the chain along the direction.
Quad Tree Encoding
This technique makes use of the principle, where a single pixel can be subdivided into any no of small pixels with same value inside it.
This principle has great advantage when it comes to storage efficiency.
Quad tree storage is a technique, where image is divided into 4 quadrants
Each quadrant is divided into 4 sub quadrants if quadrant have mixed pixels (all pixels are not same)
If quadrant is having similar pixels, the sub-divisions are not made for that quadrant.
A tree like structure is formed to store the details of divisions.
All points, where division of quadrants is happening are called as nodes.
The points where, no subdivision is happening are called as leafs.