Target detection systems

• US 5,027,413 A
• Filed: 06/13/1989
• Issued: 06/25/1991
• Est. Priority Date: 06/17/1988
• Status: Expired due to Term

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Barnard, Michael E.
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Khanna, Rohini
• Application Number:
  US07/366,102
• Time in Patent Office: 742 Days
• Field of Search: 382/1, 382/39, 358/125, 358/126, 342/192, 250/332
• US Class Current: 382/103
• CPC Class Codes: G01S 3/7864 T.V. type tracking systems
• CPC Class Codes: G01S 7/4873 by deriving and controlling a threshold value
• CPC Class Codes: G06F 18/211 Selection of the most significant subset of features
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 10/40 Extraction of image or video features
• CPC Class Codes: G06V 10/771 Feature selection, e.g. selecting representative features from a multi-dimensional feature space

Abstract

A target detection system is provided for examining picture signals of a scene which may contain a target signal (15,16) to be detected against background signals (17,18). The image is divided into sub-images and sub-stores are provided one each for the sub-images (19). The values of a set of features, such as brightness, volume or quietness, of the background and target picture signals associated with each sub-image in the respective sub-store are calculated and stored. The statistical distributions of the background and target features in each sub-store to set a target decision threshold, in feature terms, for each sub-image are calculated. The feature values of an incoming picture signal in a sub-image are compared with the decision threshold in the associated sub-store to provide a target detection signal. Each sub-image threshold relates to the image status in that particular sub-image. A lower false alarm rate with a higher detection probability is provided.

Claims

1. A target detection system comprising:

means for repetitively deriving picture signals representing an image of a scene, which picture signals may contain a target signal to be detected against background signals, in that the image is divided into a plurality of sub-images;

a plurality of sub-stores, one for each of the sub-images;

means for calculating and storing values of a set of features of background picture signals associated with each sub-image in the respective sub-store;

means for calculating and storing a statistical distribution of the background features;

means for calculating and storing values of a set of features of target signals associated with each sub-image in the respective sub-store;

means for calculating and storing a statistical distribution of the target feature;

means for comparing the statistical distribution of the background features with that of the target features to set a target decision threshold as a function of feature values for each sub-image; and

means for comparing feature values of an incoming picture signal in a sub-image with the target decision threshold in the associated sub-store to detect a target.

2. A target detection system as claimed in claim 1 characterised in that means are provided to calculate the width of the statistical distributions of the background and target features and to discard a feature as a threshold parameter if the width of the statistical distribution of that feature exceeds a predetermined amount, or if any of the feature statistics varies by more than a preset amount over a succession of scene images.

3. A target detection system as claimed in claim 1 characterised in that means are provided for applying a filter to the picture signals before the features are calculated, the filter being chosen to attenuate those signal frequencies characteristic of background signals.

4. A target detection system as claimed in claim 3 characterised in that the filter is matched to an expected target signal.