Three-dimensional object recognizing system

• US 20050264557A1
• Filed: 05/31/2005
• Published: 12/01/2005
• Est. Priority Date: 06/01/2004
• Status: Active Grant

• Current Assignee: Subaru Corp.
• Original Assignee: Fuji Heavy Industries Limited (Subaru Corp.)
• Inventors: Kise, Katsuyuki
• Granted Patent: US 7,545,975 B2
• Time in Patent Office: Days
• US Class Current: 345/419
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

A three-dimensional object recognizing system comprises a distance image generating portion for generating a distance image by using image pairs picked up by a stereoscopic camera, a grouping processing portion for grouping the distance data indicating the same three-dimensional object on the distance image, an input value setting portion for setting an area containing distance data group of grouped three-dimensional object on the distance image and also setting input values having typical distance data as elements every small area that is obtained by dividing the area by a set number of partition, a computing portion for computing output values having a pattern that responds to a previously set three-dimensional object by using a neural network that has at least the input values Xin as inputs to an input layer, and a discriminating portion for discriminating the type of the three-dimensional object based on the pattern of the output values.

Claims

1. A three-dimensional object recognizing system comprising:

a distance image generating portion for generating a distance image including three-dimensional distance data of a picked-up object, by using images picked up by an imaging portion;

a grouping processing portion for grouping the distance data indicating a same three-dimensional object on the distance image;

an input value setting portion for setting an area having a previously set predetermined profile and containing distance data group of grouped three-dimensional object on the distance image, and setting input values having typical distance data as elements every small area that is obtained by dividing the area by a set number of partition;

a computing portion for computing output values having a pattern that responds to a previously set three-dimensional object by using a neural network that has at least the input values as inputs to an input layer; and

a discriminating portion for discriminating the three-dimensional object based on the pattern of the output values computed by the computing portion.

2. A three-dimensional object recognizing system according to claim 1, wherein the input value setting portion sets a rectangular profile surrounding the distance data group on the distance image, and sets the area, each side of which circumscribes any pixel of the distance data group, by reducing the rectangular profile line by line.

3. A three-dimensional object recognizing system according to claim 1, wherein the input value setting portion sets a mean value of distance data every small area as an element of the input values.

4. A three-dimensional object recognizing system according to claim 3, wherein the input value setting portion calculates the mean value of distance data every small area at a subpixel level.

5. A three-dimensional object recognizing system according to claim 1, wherein respective weighting factors set in the neural network used by the computing portion are set by a genetic algorithm that employs a plurality of previously prepared distance images as a learning image.

6. A three-dimensional object recognizing system according to claim 1, wherein the pattern corresponding to the three-dimensional object is set as a different pattern every type of the three-dimensional object, and the discriminating portion specifies the type of the three-dimensional object based on the pattern.