Recognition algorithm for the unknown target rejection based on shape statistics obtained from orthogonal distance function

A method for using a synthetic aperture radar for acquiring a target object in a scene and computing a classification for said target object, said method comprising the steps of:

• converting a plurality of radar returns from said scene into a digital stream;
• converting said digital stream into a synthetic aperture image;
• extracting portions of said synthetic aperture image containing said target object to obtain a target region;
• applying noise compensation to said target region to obtain a noise compensated target region;
• applying anisotropic diffusion to said noise compensated target region to obtain an isolated target object;
• applying a low pass filter to said isolated target object to obtain a low pass filtered isolated target;
• applying gray scale segmentation to said low pass filtered isolated target to obtain a segmented target;
• applying object signal extraction to said segmented target to obtain a segmented image of said target subject, said segmented image having a periphery;
• selecting a plurality of random first chords across said periphery;
• measuring each of said first chords to obtain a plurality of first dimensions;
• computing for each of said first chords a second chord across said periphery thus forming a plurality of pairs of chords;
• measuring said second chords to obtain second dimensions;
• computing for each of said pairs of first chords and second chords the ratio of said first dimension to said second dimension to obtain a plurality of first values;
• computing the logarithm of said first values to obtain a plurality of logarithmic values;
• recording said first values in a first histogram;
• recording said logarithmic values in a second histogram;
• extracting a first vertex list from said first histogram;
• extracting a second vertex list from said second histogram;
• combining said first vertex list and said second vertex list to obtain a first signature, said first signature indicative of said classification of said object target.

A method for classification of a target object having a periphery comprising the steps of:

• selecting a plurality of random first chords D1 across said periphery of said target object;
• measuring each of said first chords D1 to obtain a plurality of first dimensions;
• computing for each of said first chords D1 a second chord D2 across said periphery thus forming a plurality of pairs of chords;
• measuring said second chords D2 for all pairs to obtain second dimensions;
• computing for each of said pairs of first chords D1 and second chords D2 the ratio D1/D2 of said first dimension to said second dimension to obtain a plurality of first values;
• computing the logarithm of said first values to obtain a plurality of logarithmic values;
• computing difference values by subtracting said second dimension from said first dimension for each of said pairs;
• recording said first values in a first histogram;
• recording said logarithmic values in a second histogram;
• recording said difference values in a third histogram;
• extracting a first vertex list from said first histogram;
• extracting a second vertex list from said second histogram;
• extracting a third vertex list from said third histogram;
• combining said first vertex list, and/or said second vertex list, and/or said third vertex list to obtain a signature, said signature indicative of the classification of said object target.