Matrox Electronic Systems Ltd.

Djamel Yahia Meddah

Christian Simon

Stéphane Baldo

Jon Abitz

A model-based object recognition system operates to recognize an object on a predetermined world surface within a world space. An image of the object is acquired. This image is a distorted projection of the world space. The acquired image is processed to locate one or more local features of the image, with respect to an image coordinate system of the image. These local features are mapped a world coordinate system of the world surface, and matched to a model defined in the world coordinate system. Annotations can be arranged as desired relative to the object in the world coordinate system, and then inverse-mapped into the image coordinate system for display on a monitor in conjunction with the acquired image. Because models are defined in world coordinates, and pattern matching is also performed in world coordinates, one model definition can be used by multiple independent object recognition systems.

Efficient model-based recognition of objects using a calibrated image system

A method for locating an object in an image is provided for an object that exhibits, in at least one direction, foreground pixel runs and background pixel runs consistent with at least characteristic of the object. In one embodiment, the method includes the acts of defining a plurality of regions in the image, associating a metric with each of the regions, finding pixel runs in the image, and identifying at least one pixel run in one of the plurality of regions where the at least one pixel run is consistent with the at least one characteristic of the object. According to one embodiment, the method includes the acts of adjusting the metric associated with the region where the at least one pixel run is identified and identifying, from among the plurality of regions, a region where the object is located based on a metric determined for a plurality of regions. In various embodiments, the object includes a repetitive pattern such as those found in barcodes and character strings.

Methods and apparatus for locating an object in an image

Model-based recognition of objects using a calibrated image system

A method of pattern matching for recognition of objects within an image, uses a model defined by a set of one or more model primitives representative of large scale structures of an archetype. The image is analyzed to derive at least one target primitive representative of a large scale structure of the image. A basis is derived from at least one of the target primitives, and each of the target primitives represented as invariants in the derived basis. For each target primitive, any model primitives that at least partially match the target primitive are identified. Each identified model primitive is then processed to calculate a figure of merit indicative of a degree of correspondence between the target primitive and the model primitive. In some embodiments, the figure of merit is an Orthogonal projection between the target and model primitives, which is and accumulated for each model basis.

Geometric hashing method for model-based recognition of an object

A method for recognizing an object in a target image using model primitives comprising an additive primitive and a subtractive primitive; weights are assigned to the additive and subtractive primitives; a target primitive is derived for the object; associations are determined between the target primitive and the model primitives; a similarity score is computed for the target primitive with respect to the model primitives; the similarity score is increased for each association between the target primitive and the additive primitive and decreased for each association between the target primitive and the subtractive primitive; the weights determine an amount by which the similarity score is increased or decreased for each of the associations.

Growing community of inventors

Montreal, Canada

Djamel Yahia Meddah