Introduction to the IDRISI for Windows 2 system
GETTING TO
KNOW IDRISI
File formats in IDRISI: Image
files, the rasterdata
[1] images, [2] vectors, [3] values files, [4] reference files, [5] composition files, [6] others
| learn about the various file types in IDRISI. Shown here are the most important ones only. Others will appear as the tutorial passes by specific modules which need their own file types |
Getting to know IDRISI
File maintenance - File formats - Using DOCUMENT - Display system - Palette/Symbol
Workshop
Let us imagine a very simple image:
| row col |
0 | 1 | 2 | 3 | 4 |
|---|---|---|---|---|---|
| 0 | 22 | 22 | 18 | 18 | 18 |
| 1 | 15 | 15 | 18 | 16 | 16 |
| 2 | 11 | 15 | 15 | 18 | 16 |
| 3 | 11 | 15 | 12 | 12 | 12 |
5 columns, 4 rows. The values may represent some code for land usage. IDRISI is starting in the upper-left corner (row 0/column 0), then advances column by column and row by row. In the simplest format - ASCII - the cellvalues are stored one in each line:
22
22
18
18
18
15
15
(...)
11
15
12
12
12
Commonly the images are stored binary, one value after the other. Depending on the datatype a value occupies more or less of memory. Simple RLC (run length compression) is supported as packed binary (i.e., the cellvalue is followed by the number of occurences in the series):
- not packed: 22 22 18 18 18 15 15 18 16 16 11 15 15 18 16 11 15 12 12 12
- packed: 22 2 18 3 15 2 18 1 16 2 11 1 15 2 18 1 16 1 11 1 15 1 12 3
Unfortunately our image is not well suited for compression. The compressed image requires 4 additional bytes! Not the best example to demonstrate. But images with large areas of same values can show packing ratios up to 1:100 and more in case of significant homogeneous areas.
The table below shows IDRISI's datatypes (except of compression they apply to vectorfiles as well):
| memory required | range | compression | |
|---|---|---|---|
| byte | 1 byte | 0 to 255 | yes |
| integer | 2 bytes | -32768 to +32767 | yes |
| real | IEEE 4 bytes | ±1*10 38, 7 significant figures precision | no |
Evidently the image 'does not know' about its dimensions or about the area one cell covers in real nor the units it represents. So we need a 'header file', the documentation file, which corresponds with a image.
