US 5,757,977 AGranted

Fuzzy logic filter for reducing noise and sharpening edges of digital image signals

AssigneeSGS-Thomson Microelectronics S.A.

InventorsMassimo Mancuso, Rinaldo Poluzzi, Gianguido Rizzotto

Life of this patent

Estimating the term…

Loading patent…

Abstract

A filter achieving noise reduction and digital signal image edge enhancement includes two noise reduction circuits designed to enhance an image edge. The noise reduction circuits comprise both comparison units, which receive separate digital signals of an image and an inferential circuit connected to said comparison elements. Each inferential circuit uses a fuzzy logic unit designed to define activation levels based on signals generated by the comparison elements. The filter also includes a noise detection circuit and an image edge detection circuit both connected to the noise reduction circuit and designed to perform operations in accordance with fuzzy logic rules on the basis of the activation levels defined in the inferential circuits. The filter also includes a noise reduction circuit connected to the noise detection circuit and is designed to filter the digital image signals on the basis of the operations performed by the noise detection circuit. The filter additionally includes an image edge enhancement circuit connected to the noise reduction circuit and to the image edge detection circuit and is designed to perform on filtered digital image signals image edge enhancement on the basis of the operations performed by the image edge detection circuit.

Claims17

Dependent claims

A fuzzy logic filter for digital image signals, comprising:

a first and second processing circuit each having a first and second comparison element each for accepting a comparison digital signal on a first input, a measuring digital signal on a second input, and providing a differential signal on an output, and each processing circuit having an inferential circuit with a first and second input coupled to the output of each comparison element, respectively, each inferential circuit for defining activation levels and providing them to an output according to fuzzy logic rules based on the differential signal inputs;
a noise detection circuit having a first input coupled to the output of the inferential circuit of the first processing circuit, and a second input coupled to the output of the inferential circuit of the second processing circuit, the noise detection circuit for providing noise level signals to an output, the noise level signals produced according to fuzzy logic rules based on the activation level inputs;
a noise reduction circuit having a first input coupled to the output of the noise detection circuit, and a second input coupled to digital image signals, the noise reduction circuit for providing filtered image signals to an output according to fuzzy logic rules, the amount of filtering based on the input noise level signals;
an edge detection circuit having inputs connected to the respective output of the inferential circuit of both the first and second processing circuit, the edge detection circuit for providing edge probability signals to a pair of outputs according to fuzzy logic based on the activation level inputs;
an edge enhancement circuit having a first and second input coupled to the respective outputs of the edge detection circuit, and having a third input coupled to the output of the noise reduction circuit, the edge enhancement circuit for producing, to an output, enhanced digital image signals according to fuzzy logic rules based on the edge probability signal inputs and the digital image signal input.

2 more elements

A fuzzy logic filter for digital image signals having a first edge and a second edge, comprising:

a first processing circuit having as input a first set of the digital image signals including a target signal, the first processing circuit for inferring according to fuzzy logic a first activation level that represents how reliably the target signal may belong to the first edge of the digital image signals;
a second processing circuit having as input a second set of the digital image signals including the target signal, the second processing circuit for inferring according to fuzzy logic a second activation level that represents how reliably the target signal may belong to the second edge of the digital image signals;
a noise detection circuit connected to the first and second processing circuits, the noise detection circuit for determining a noise level of the target signal based on the inferred first and second activation levels;
a noise reduction circuit connected to the noise detection circuit, the noise reduction circuit for filtering the target signal based on the determined noise level, the amount of filtering being greater when the determined noise level indicates a greater noise level;
an image edge detection circuit connected to the first and second processing circuits, the image edge detection circuit for producing a first reliability level that represents a likelihood that the target signal belongs to the first edge of the digital image signals rather than the second edge, and producing a second reliability level that represents a likelihood that the target signal belongs to the second edge of the digital image signals rather than the first edge;
and
an image edge sharpening circuit connected to the image edge detection circuit and the noise reduction circuit, the image edge sharpening circuit for selecting the first edge when the first and second reliability levels indicate that the target signal more likely belongs to the first edge than the second edge and selecting the second edge when the first and second reliability levels indicate that the target signal more likely belongs to the second edge than the first edge, and sharpening the selected edge on the filtered target signal.

4 more elements

Dependencies are inferred from the claim text.