OBJECT DETECTION DEVICE

• US 20120076361A1
• Filed: 06/02/2010
• Published: 03/29/2012
• Est. Priority Date: 06/03/2009
• Status: Active Grant

• Current Assignee: Chubu University Educational Foundation
• Original Assignee: Chubu University Educational Foundation
• Inventors: Fujiyoshi, Hironobu
• Granted Patent: US 8,611,604 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G01S 17/894 3D imaging with simultaneous measurement of time-of-flight at a 2D array of receiver pixels, e.g. time-of-flight cameras or flash lidar
• CPC Class Codes: G01S 7/4802 using analysis of echo signal for target characterisation; Target signature; Target cross-section
• CPC Class Codes: G06T 2207/10028 Range image; Depth image; 3D point clouds
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
• CPC Class Codes: G06V 20/64 Three-dimensional objects
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

A depth histogram is created for each of a plurality of local regions of the depth image by grouping, according to specified depths, the depth information for the individual pixels that are contained in the local regions. A degree of similarity between two of the depth histograms for two of the local regions at different positions in the depth image is calculated as a feature. A depth image for training that has a high degree of certainty is defined as a positive example, a depth image for training that has a low degree of certainty is defined as a negative example, a classifier that is suitable for classifying the positive example and the negative example is constructed, and an object that is a target of detection is detected in the depth image, using the classifier and based on the feature.

Claims

1. An object detection device, comprising:

a depth image acquisition means that acquires depth information that expresses distances to an object of image capture and acquires a depth image in which information for individual pixels is the depth information;

a depth histogram creation means that creates a depth histogram for each of a plurality of local regions of the depth image by grouping, according to specified depths, the depth information for the individual pixels that are contained in the local regions;

a feature calculation means that calculates, as a feature, a degree of similarity between two of the depth histograms for two of the local regions at different positions in the depth image; and

a classification means that defines a depth image for training that has a high degree of certainty as a positive example, defines a depth image for training that has a low degree of certainty as a negative example, constructs a classifier that is suitable for classifying the positive example and the negative example, and detects in the depth image, using the classifier and based on the feature, an object that is a target of detection.

2. The object detection device according to claim 1,

wherein the classification means includes

an occlusion determination means that, in a case where the depth information for one of the pixels that are contained in one of the local regions indicates a depth that is less than a specified threshold value, determines that the pixel represents an occlusion,

an occlusion rate calculation means that calculates an occlusion rate that is a ratio of the pixels in the local region that have been determined to represent an occlusion, and

a classifier construction means that constructs the classifier using the occlusion rate.

3. The object detection device according to claim 1,

wherein the depth image acquisition means is in a fixed position in a three-dimensional space, acquires a full-space depth image that contains the depth image, and acquires from the full-space depth image, by raster scanning the three-dimensional space, the depth image that corresponds to the object.

4. The object detection device according to claim 1,

wherein the depth image acquisition means is a time-of-flight camera.

5. The object detection device according to claim 2,

wherein the depth image acquisition means is in a fixed position in a three-dimensional space, acquires a full-space depth image that contains the depth image, and acquires from the full-space depth image, by raster scanning the three-dimensional space, the depth image that corresponds to the object.

6. The object detection device according to claim 2,

wherein the depth image acquisition means is a time-of-flight camera.

7. The object detection device according to claim 3,

wherein the depth image acquisition means is a time-of-flight camera.

8. The object detection device according to claim 5,

wherein the depth image acquisition means is a time-of-flight camera.