Accession Number:

ADA106688

Title:

Characteristics of Images Derived from Factorable Polynomials

Descriptive Note:

Final technical rept. Feb-Dec 1980

Corporate Author:

SCRIPPS INSTITUTION OF OCEANOGRAPHY LA JOLLA CA VISIBILITY LAB

Personal Author(s):

• McGlamery, B. L.

Report Date:

1981-09-01

Pagination or Media Count:

46.0

Abstract:

Image reconstruction techniques exist which use as the input the autocorrelation function of the image. The uniqueness of the reconstructed image has been questioned. This report considers a class of images which are ambiguous in that many different images can be generated which have the same autocorrelation function. It is shown that two dimensional images generated from factorable polynomials which have unity power factors are basically uninteresting images. However, by using non-unity power factors, images with structure results. This structure varies significantly among images with the same autocorrelation function. Finally, it is pointed out that this factoring process is equivalent to convolution of the image with a pair of delta functions, leading to the ability of generating ambiguous images from functions that are not completely factorable. Examples are shown. The conclusion of the work is that caution must be exercised in interpretation of images reconstructed from autocorrelation functions.

Descriptors:

• *IMAGE PROCESSING
• *IMAGE RESTORATION
• *AUTOCORRELATION
• POLYNOMIALS
• TABLES(DATA)
• FREQUENCY DOMAIN
• FACTOR ANALYSIS
• TWO DIMENSIONAL
• SPATIAL DISTRIBUTION
• FOURIER TRANSFORMATION
• BRIGHTNESS

Subject Categories:

• Statistics and Probability
• Human Factors Engineering and Man Machine Systems

Distribution Statement:

APPROVED FOR PUBLIC RELEASE