ES 555 image interpretation

Image Interpretation
ES 555 Small Format
Aerial Photography

Table of Contents
Basic features Nature of resolution
Rating criteria Analog vs. digital
Related sites References

**Table of Contents**
Basic features	Nature of resolution
Rating criteria	Analog vs. digital
Related sites	References

Basic features

Visual interpretation of aerial photographs is based on recognition of objects, as seen from above, in pictorial format. This visual recognition often takes place without any conscious effort by the interpreter. Nonetheless, there are several basic features that can aid in visual examination and interpretation of airphotos.

Tone or color -- Grey tone (b/w) or color are how we recognize and distinguish objects. The "color" of an object helps to separate it from other features in the scene, especially for features with high contrast. Color may be an important clue to the object's identity--water, soil, vegetation, buildings, etc.
Shape -- Natural shapes tend to follow the "lay of the land" or be irregular. Cultural (man-made) shapes, on the other hand, are often geometric in nature with straight lines, sharp angles, and regular forms.
Size -- Absolute size is an important clue that depends on scale of the photograph. Some object sizes are common experience, for example a car, tree, etc. But many objects seen on airphotos are not so obvious. Always check photo scale for a guide to object size.
Pattern -- The spatial arrangement of individual objects may create a distinctive pattern. This is most apparent for cultural features, for example city street grids, airport runways, agricultural fields, etc. Patterns in the natural environment may also be noticeable in some cases, for example bedrock fractures, drainage networks, etc.
Texture -- This refers to grouped objects that are too small or too close together to create distinctive patterns. Examples include tree crowns in a forest canopy, individual plants in an agricultural field, cars in a traffic jam, etc. The difference between texture and pattern is largely determined by photo scale.
Shadow -- Shadows can be useful clues for identifying objects, as seen from above. Trees, buildings, animals, bridges and towers are examples of features that cast distinctive shadows. Shadows on the landscape also help with depth perception. Photographs without shadows often seem "blank" and difficult to interpret; however, too much shadowing may obscure features of interest.
Context -- The association of objects is often important for aiding interpretation. Note landcover and landuse as clues to help identify related features in the scene, and refer to existing maps or data for additional information. This is often a matter of "common sense" for image interpretation.

Nature of resolution

The term resolution is used in different ways in remote sensing. In general, it refers to the smallest identifiable feature within an image. For a digital image, this means an individual cell or pixel (picture element). For conventional film, the number of resolvable lines per inch (or mm) is a measure of resolution. However, these quantitative approachs ignore the fact that size is not the only element necessary for identifying objects in an image. As noted above shape, pattern, color, contrast, texture, shadow, and context are other elements that aid in feature recognition.

Rule of thumb: Positive recognition of objects in aerial photographs requires a ground resolution three to five times smaller than the object size (Hall 1997).

For example, to positively identify an automobile (width 2 m), the camera/image system needs to achieve a ground resolution between 0.67 and 0.4 meters. With these limitations in mind, a perceptual measure of image quality or interpretability has been developed. This measure is the National Imagery Interpretability Rating Scale--NIIRS (Leachtenauer et al. 1997). An image rating depends on the most difficult interpretation task that can be performed, which indicates the level of interpretability that can be achieved. The following listing provides the civilian NIIRS criteria.

Civilian NIIRS criteria

Based on (Leachtenauer et al. 1997).

Rating Level 0

Interpretability of imagery precluded by obscuration, degradation, or very poor resolution.

Rating Level 1

Distinguish between major landuse classes--urban, agricultural, forest, water, barren, etc.
Detect medium-sized port facility.
Distinguish between runways and taxiways at a large airport.
Identify regional drainage patterns--dendritic, trellis, etc.

Rating Level 2

Identify large (> ¼-section) center-pivot irrigated fields.
Detect large buildings--hospital, factory, etc.
Identify road patterns--highway interchange, road network, etc.
Detect ice-breaker tracks in sea ice.
Detect wake of large (> 100 m) ship.

Rating Level 3

Detect large area (> ¼-section) contour plowing.
Detect individual houses in residential neighborhoods.
Detect trains or strings of railroad cars on tracks.
Identify inland waterways navigable by barges.
Distinguish between natural forest and orchards.

Rating Level 4

Identify farm buildings as barns, silos, or residences.
Count tracks along railroad right-of-way or in train yards.
Detect basketball court, tennis court, volleyball court in urban area.
Identify individual tracks, control towers or switching points in railroad yard.
Detect jeep trails through grassland.

Rating Level 5

Identify Christmas-tree plantation.
Recognize individual train cars by type--box, flat, tank, etc.
Detect open bay doors of vehicle storage buildings.
Identify large tents within camping areas.
Distinguish between stands of deciduous and coniferous forest during leaf-off season.
Detect large animals (elephant, rhinoceros, bison, etc.) in grassland.

Rating Level 6

Detect narcotics intercropping based on texture.
Distinguish between row crops (corn, milo, soybeans) and small grains (wheat, barley, oats).
Identify passenger vehicles as sedans, station wagons, etc.
Identify individual utility poles in residential neighborhoods.
Detect foot trails in barren areas.

Rating Level 7

Identify individual mature plants in a field of known row crops--cotton, corn, etc.
Recognize individual railroad ties.
Detect individual steps on stairways.
Detect stumps and rocks in forest clearings or meadows.

Rating Level 8

Count individual baby farm animals--pigs, sheep, etc.
Identify a survey benchmark set in a paved surface.
Identify body details or license plate on passenger car or truck.
Recognize individual pine seedlings or individual water lilies on a pond.
Detect individual bricks in a building.

Rating Level 9

Identify individual seed heads on grain crop--wheat, oats, barley, etc.
Recognize individual barbs on a barbed-wire fence.
Detect individual spikes in railroad ties.
Identify individual leaves on a tree.
Identify an ear tag on large animals--deer, elk, cattle, etc.

Manned-space photography, as practiced on space-shuttle missions, provides images of the Earth's surface for rating levels 1-3. Rating levels 4-6 are typically attained by conventional aerial photography, and sometimes level 7 is possible. However, the highest rating levels (7-9) are generally not available for civilian use. These highest NIIRS rating levels (7-8) may be acquired, in many cases, with SFAP from low-height platforms.

Analog vs. digital imagery

Conventional photography with film is called analog imagery, as the photographic emulsion records continuous variations in light intensity. Digital cameras, on the other hand, record light intensity in discrete intervals. High-resolution digital cameras now achieve the same spatial resolution as conventional 35-mm films, although medium- and large-format films still have a spatial resolution advantage. This situation is likely to change quickly, as digital resolution continues to improve rapidly.

The conversion of analog (film) photographs into digital format by scanning carries a potential disadvantage--resolution of the digital image may be degraded both spatially and in terms of tone/color range. The amount of degradation can be estimated by comparison of NIIRS rating levels for the original photograph and scanned digital image. On this basis, the scanning technique may be adjusted in order to preserve a necessary level of interpretability for a given project or application of the digital imagery.

Related sites

Civil NIIRS reference guide from the Federation of American Scientists.
Multispectral imagery interpretability rating scale from the Federation of American Scientists.

References

Hall, R.C. 1997. Post war strategic reconnaissance and the genesis of project Corona. In R.A. McDonald (ed.), Corona: Between the Sun & the Earth, The first NRO reconnaissance eye in space, p. 25-58. American Society Photogrammetry and Remote Sensing.
Leachtenauer, J.C., Daniel, K. and Vogl, T.P. 1997. Digitizing Corona imagery: Quality vs. cost. In R.A. McDonald (ed.), Corona: Between the Sun & the Earth, The first NRO reconnaissance eye in space, p. 189-203. American Society Photogrammetry and Remote Sensing.