Home Articles Image Interpretation of Remote Sensing data

Image Interpretation of Remote Sensing data

Dr R Sivakumar
Senior Faculty, GIS Institute
[email protected]
Remote Sensing is the science and art of obtaining information about an object/phenomena or area through the analysis of data acquired by a device that is not in contact with the object under investigation. This technology first emerged in 1840 and pictures were taken by balloon. Subsequently, Cameras were mounted in airplane for military survey in the first world war for reconnaissance survey.

This technology includes both Satellite and aerial remote sensing. The basis source for this technology is electromagnetic radiation and this energy from the sun reaches the earth surface and again reflected or transmitted or absorbed by the objects which is collected by the satellite sensors or recorded in the photographic film. The product from the aerial camera is called �photograph� and the term �image� is used for any pictorial representation of image data. The reflectance/remittance/absorption of energy by an object forms the base for the brightness or darkness in an image or photographs. This is further interpreted for the identification of the features. The earlier stage aerial and satellite data were in black & white and subsequent advancement in sensor development and colour film leads the generation of colour photographs and images. The significant advance in sensor technology stemmed from subdividing spectral ranges of electromagnetic radiation into several bands allowing sensors in several bands to form multispectral images In general there are three different types of data products namely black and white photograph or panchromatic image (single band), normal colour and false colour composite (Multispectral ). Single band image displays as a gray scale but combination of three bands at a time generates colour composite images (Fig.1).

Interpretation is the processes of detection, identification, description and assessment of significant of an object and pattern imaged. The method of interpretation may be either visual or digital or combination of both. Both the interpretation techniques have merits and demerits and even after the digital analysis the output are also visually analysed.

Green band (1) Red band (2) Near IR band (3) False colour
Composite (123)

Figure 1: Combination of 3 bands generates colour composite images.
Source: www.sci-ctr.edu.sg/ssc/publication/remotesense/opt_int.html

The ability of human to identify an object through the data content in an image/photo by combining several elements of interpretation. There are two types of extraction of information from the images/photographs namely;

  1. Interpretation of data by visual analysis,
  2. Semi automatic processing by computer followed by visual analysis like generation of vector layer from raster image through onscreen digitisation and DTM/DEM generation. Similarly interpretation of aerial photographs through 3D generation through visual studies. In general analog format in remote sensing data is being used in visual interpretation. This involves the systematic examination of data, studying existing maps, collection of field information and works at various levels of complexity. The analysis depends upon the individual perception, and experience of the interpreter, nature of the object, quality of the data, scale, combination of special bands etc.
    The entire process of visual interpretation can be divided into following few steps namely detection of an object, interpretation, recognition and identification, analysis, classification, deduction and idealisation and based on this identifying an object conclusion. Hence interpretation is the combined result of identification of feature through photo recognition elements, field verification and preparation of final thematic maps. It also requires the process of observation coupled with imagination and great deal of patience.

Basic elements of interpretation
The interpretation of satellite imagery and aerial photographs involves the study of various basic characters of an object with reference to spectral bands which is useful in visual analysis. The basic elements are shape, size, pattern, tone, texture, shadows, location, association and resolution.

Shape: The external form, outline or configuration of the object. This includes natural features (Example: Yamuna River), in Delhi Man Made feature (Example : Nehru Stadium, Delhi.
Size : This property depends on the scale and resolution of the image/photo. Smaller feature will be easily indented in large scale image/photo.
Pattern: Spatial arrangement of an object into distinctive recurring forms: This can be easily explained through the pattern of a road and railway line. Eventhough both looks linear, major roads associated with steep curves and many intersection with minor road.
Shadow: Indicates the outline of an object and its length which is useful is measuring the height of an object. The shadow effect in Radar images is due to look angle and slope of the terrain. Taller features cast larger shadows than shorter features.
Tone: Refers to the colour or relative brightness of an object. The tonal variation is due to the reflection, emittance, transmission or absorption character of an objects. This may vary from one object to another and also changes with reference to different bands. In General smooth surface tends to have high reflectance, rougher surface less reflectance. This phenomenon can be easily explained through Infrared and Radar imagery .
Infrared imagery: Healthy vegetation reflects Infrared radiation much more stronger than green energy and appears very bright in the image. A simple example is the appearance of light tone by vegetation species and dark tone by water. Particularly in thermal infrared images the brightness tone represents warmest temperature and darkness represent coolest temperature. The image (Fig2) illustrates daytime and night time thermal data. The changes in kinetic water temperature cause for the tonal changes. Hence time is also to be taken consideration before interpretation
Radar Imagery : Smooth surfaces reflect highly and area blocked from radar signal and appear dark. Bridges and cities show very bright tone, on the contrary calm water, pavement and dry lake beds appears very dark tone.
Texture: The frequency of tonal change. It creaks a visual impression of surface roughness or smoothness of objects. This property depends upon the size, shape, pattern and shadow:
Location Site : The relationship of feature to the surrounding features provides clues to words its identity. Example: certain tree species words associated with high altitude areas
Resolution: It depends upon the photographic/imaging device namely cameras or sensors. This includes of spectral and spatial resolutions. The spectral resolution helps in identifying the feature in specific spectral bands. The high spatial resolutions imagery/photographs is useful in identifying small objects.
Association: Occurrence of features in relation to others.

Hence, careful examination has to be done to identify the features in the imagery combined with field information.

Issues in interpretation.

  • Unfamiliar scale and resolutions.
  • Lack of understanding of physics of Remote sensing.
  • Understanding proper spectral character of each object
  • visually interpret 3 layers of information at a time.

Daytime thermal image Night time thermal image

Figure 2: Daytime and night time thermal data.
Source: Lillesand TM and Kiefer RW (2000), Remote Sensing and Image Interpretation.

Atlanta Delhi

Figure 3: IKONOS image Credit: www.spaceimaging.com
Success of interpretation

  • Training and Experience of the interpreter
  • Quality of photo/Images
  • Local knowledge of the study area.

Advantages in visual Interpretation

  • Simple method
  • Inexpensive equipment
  • Uses brightness and spatial content of the image
  • Subjective and Qualitative
  • Concrete

Advantages of digital image processing:

  • Cost-effective for large geographic areas
  • Cost-effective for repetitive interpretations
  • Cost-effective for standard image formats
  • Consistent results
  • Simultaneous interpretations of several channels
  • Complex interpretation algorithms possible
  • Speed may be an advantage
  • Explore alternatives
  • Compatible with other digital data

Disadvantages in digital processing:

  • Expensive for small areas
  • Expensive for one-time interpretations
  • Start-up costs may be high
  • Requires elaborate, single-purpose equipment
  • Accuracy may be difficult to evaluate
  • Requires standard image formats
  • Data may be expensive, or not available
  • Preprocessing may be required
  • May require large support staff

High Resolution images
The spatial resolution of an image /photograph is one of the main element to decide the scale of map and classification level of information. The advances in sensor development led the acquiring of high resolution imagery which helps planners, and professionals for making large scale maps, better planning and monitoring. The panchromatic data of IKONOS (Fig.3) with the resolution of 1mt/4mt is being used widely for large scale mapping. These high resolution imagery can help in fast visual interpretation by human eyes, the best image processor of all and even small objects be easily detected. Hence the human brain can identify an object by understanding, analyzing the context . The chances of misinterpretation of object is very less in visual analysis because of the expertise, experience and local knowledge of the field.