Device and method for classifying objects in an environmentally adaptive manner
First Claim
1. A device for the automatic classification of an object by means of Doppler-broadened radar echo signals reflected by the object and received by a radar receiver, the device including a neural network (NET) with an input layer (IL) of input nodes (IN1, . . . , IN57) for inputting features (M) of the Doppler-broadened radar echo signals, and an output layer (OL) of output nodes (ON1, ON2, ON3) each corresponding to a class in a predetermined group of classes into one of which the object can be allocated, characterized in that the neural network (NET) has at least one additional input node (ZN1, ZN2) into which control information (SI) can be entered, for causing the neural network (NET) to adapt to an external influence factor.
2 Assignments
0 Petitions
Accused Products
Abstract
Neural networks are used to classify objects automatically by means of Doppler-broadened radar echo signals. The classification device KK contains a neural network (NET, NET2) which has an input layer (IL) of input nodes (IN1, . . . , IN57) for features (M) of the Doppler-broadened radar echo signals, and an output layer (OL) of output nodes (ON1, ON2, ON3) for predetermined classes to which the objects can be allocated. The neural network (NET, NET2) is adapted to the external conditions prevailing at the time of the classification operation. The adaptation takes place either via accessible input nodes (ZN1, ZN2) into which control information (SI) can be entered, and which cause the neural network (NET) to adapt to one or to several external influence factors, or via a selection device (SEL) which, from the parameters (P1, . . . , P4) of several neural networks stored in a memory (MEM), which are trained with training data under respectively different conditions of external influence factors, selects the one most similar to the prevailing conditions.
37 Citations
11 Claims
- 1. A device for the automatic classification of an object by means of Doppler-broadened radar echo signals reflected by the object and received by a radar receiver, the device including a neural network (NET) with an input layer (IL) of input nodes (IN1, . . . , IN57) for inputting features (M) of the Doppler-broadened radar echo signals, and an output layer (OL) of output nodes (ON1, ON2, ON3) each corresponding to a class in a predetermined group of classes into one of which the object can be allocated, characterized in that the neural network (NET) has at least one additional input node (ZN1, ZN2) into which control information (SI) can be entered, for causing the neural network (NET) to adapt to an external influence factor.
-
9. A device for automatically classifying objects by means of Doppler-broadened radar echo signals, the device including a neural network (NET) having an input layer (IL) of input nodes (IN1, . . . , IN57) each for inputting a feature (M) of the Doppler-broadened radar echo signals, and an output layer (OL) of output nodes (ON1, ON2, ON3) for allocating the object to a class in a predetermined set of classes, characterized in that the device further comprises:
-
a memory (MEM) for storing a plurality of parameter sets for use with the neural network, each parameter set determined from training under different sets of conditions of external influence factors, each parameter set corresponding to a set of conditions of external influence factors, and a selection device (SEL) for selecting, as a function of external influence factors, one parameter set from the plurality of parameter sets.
-
-
10. A method of automatically classifying an object by means of Doppler-broadened radar echo signals reflected from the object, in which
features (M) are determined from the Doppler-broadened radar echo signals and are presented to a neural network (NET) at input nodes (IN1, . . . , IN57) of an input layer (IL) of the neural network, and as a response to the features (M) presented to the neural network (NET), the neural network activates output nodes (ON1, ON2, ON3) to classify the object, characterized in that an adaptation of the neural network (NET) to at least one external influence factor is carried out, by determining a control information (SI) value corresponding to the at least one external influence factor, which is entered into an additional input node (ZN1, ZN2) of the neural network (NET).
Specification