Browse Prior Art Database

3D Fourier Transform Technique for Motion Detection

IP.com Disclosure Number: IPCOM000088320D
Original Publication Date: 1977-May-01
Included in the Prior Art Database: 2005-Mar-04
Document File: 2 page(s) / 31K

Publishing Venue

IBM

Related People

Winter, EM: AUTHOR

Abstract

This technique allows the detection of moving objects with very low signal-to-noise ratios using an electro-optical imaging sensor system. The method makes maximum usage of a priori motion data while not requirining accurate position data. Image data samples are taken periodically. A Fourier transform (FT) of this sampled data is taken with image plane coordinates x and y along with time, comprising the three dimensions. The transformed data is multiplied by a similarly transformed frequency template of the expected object. The resulting product is transformed back to space-time coordinates. A Functional Flow of the algorithm is shown in the figure.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 78% of the total text.

Page 1 of 2

3D Fourier Transform Technique for Motion Detection

This technique allows the detection of moving objects with very low signal-to-noise ratios using an electro-optical imaging sensor system. The method makes maximum usage of a priori motion data while not requirining accurate position data. Image data samples are taken periodically. A Fourier transform (FT) of this sampled data is taken with image plane coordinates x and y along with time, comprising the three dimensions. The transformed data is multiplied by a similarly transformed frequency template of the expected object.

The resulting product is transformed back to space-time coordinates.

A Functional Flow of the algorithm is shown in the figure.

The 3D FT algorithm is an extension of digital filtering techniques to the three dimensional space-time domain. A space-time image, i(x, y, t), produced by an electro-optical image sensor, such as a charge-coupled device (CCD), is transformed into an equivalent frequency representation.

A three-dimensional target matched filter is formed by using a priori data on velocity and direction. The space-time history is represented by a real array of intensity values, I(F) (x, y, t). The 3D transform is formed from this array. I(F) (f(x), f(y), f(t)) = T [I(F) (x'y,t)] where T represents a transform operation. A filter can then be formed by taking the complex conjugate.

The filtering operation is performed by taking the transform of the observed data, (f(x), f(y), f(t)) = T...