Object detection device, object detection method, and object detection program

• US 9,152,887 B2
• Filed: 11/08/2012
• Issued: 10/06/2015
• Est. Priority Date: 06/02/2011
• Status: Active Grant

• Current Assignee: Panasonic Automotive Systems Company of America (Panasonic Holdings Corporation)
• Original Assignee: Panasonic Intellectual Property Management Co., Ltd. (Panasonic Holdings Corporation)
• Inventors: Muramatsu, Hirokazu
• Primary Examiner(s): ELAHI, SHAN E
• Application Number:
  US13/672,002
• Publication Number:
  US 20130070096A1
• Time in Patent Office: 1,062 Days
• Field of Search: 348/148
• US Class Current: 1/1
• CPC Class Codes: G06F 18/253 of extracted features
• CPC Class Codes: G06V 10/7515 Shifting the patterns to accommodate for positional errors
• CPC Class Codes: G06V 10/806 of extracted features
• CPC Class Codes: G06V 20/58 Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

An object detection device (10) is provided with: a video image converting section (20) converting an input video image in which surroundings of a vehicle are shot, to a characteristics video image into which image characteristics have been extracted from the input video image; a video images-classified-by-distance extracting/composing section (30) extracting areas which differ according to distances from the characteristics video image on the basis of the distance from a vehicle and composing the areas; a first object detecting section (40) scanning a composite video image to perform first object detection processing; an object-candidate position specifying section (50) determining an object-candidate position from a result of the first object detection processing; a second object detecting section (60) performing second object detection processing for the object-candidate position; and an object position identifying section (70) identifying an object position from a result of the second object detecting section.

Claims

1. An object detection device detecting an object near a vehicle from an input video image, the input video image being a video image of surroundings of the vehicle shot from the vehicle, the device comprising:

a video image converting section which converts the input video image to a characteristics video image in which image characteristics have been extracted from the input video image;

a video image processing section which extracts areas which differ according to distances from the characteristics video image as video images classified by distance, on the basis of the distance from the vehicle, and composes a composite video image using the video images classified by distance, wherein the composite video image is a composite of said video images classified by distance;

a first object detecting section which performs first object detection processing of scanning the composite video image composed by the video image processing section to determine a position on the composite video image of an object detected from the composite video image;

an object-candidate position specifying section which determines an object-candidate position where there is a possibility of existence of the object, in the video images classified by distance, on the basis of the position on the composite video image of the object detected by the first object detecting section;

a second object detecting section which performs, for the object-candidate position determined by the object-candidate position specifying section, second object detection processing for identifying corresponding object positions in the video images classified by distance; and

an object position identifying section which identifies an object position in the input video image on the basis of the object positions determined by the second object detecting section.

2. The object detection device according to claim 1, wherein the video image converting section extracts edge characteristics as the image characteristics.

3. The object detection device according to claim 1, wherein the video image processing section changes the size of an area extracted from the characteristics video image on the basis of the distance from the vehicle.

4. The object detection device according to claim 1, wherein the video image processing section enlarges or reduces the video images according to distances so that the vertical sizes of all the video images classified by distance are equal to one another.

5. The object detection device according to claim 1, wherein the video image processing section adjusts the positions of the horizontal center axes and vertical center axes of the video images classified by distance so that they are overlapped, and composes the composite video image.

6. The object detection device according to claim 1, wherein the video image processing section composes the composite video image by distance by α

blending.

7. The object detection device according to claim 1, wherein the video image processing section partially adjusts a composite coefficient of α

blending and composes the composite video image, according to the number of composed video images classified by distance.

8. The object detection device according to claim 1, wherein the video image processing section adjusts the composite coefficient of α

0 blending and composes the composite video image, according to the numbers of image characteristics included in the video images classified by distance.

9. The object detection device according to claim 1, wherein the first object detecting section sets the vertical size of a detection target object to be equal to the vertical size of the composite video image in the first object detection processing.

10. The object detection device according to claim 1, wherein the first object detecting section performs object detection processing by performing scanning on the composite video image only in a horizontal direction.

11. The object detection device according to claim 1, wherein the first object detecting section comprises a composite video image processing section cutting out an area in contact with the bottom end of the composite video image and generating an enlarged composite video image obtained by enlarging the vertical size of the cut-out area so that the vertical size is equal to the vertical size of the composite video image.

12. The object detection device according to claim 11, wherein the first object detecting section performs the object detection processing for the enlarged composite video image.

13. The object detection device according to claim 12, wherein the first object detecting section performs the object detection processing by performing scanning on the enlarged composite video image only in a horizontal direction.

14. The object detection device according to claim 1, wherein the first object detecting section is capable of adjusting a detection judgment criterion for object detection separately from the second object detecting section, and the detection criterion is adjusted so that the first object detecting section judges an object to be an object more easily than the second object detecting section.

15. The object detection device according to claim 1, wherein the first object detecting section performs the first object detection processing only on parts of the composite video image where the image characteristics exist as targets.

16. The object detection device according to claim 1, wherein the object position identifying section determines the distance from the vehicle to a detected object on the basis of the object positions on the video images classified by distance detected by the second object detecting section.

17. An object detection method for detecting an object near a vehicle from an input video image, the input video image being a video image of surroundings of the vehicle shot from the vehicle, the method comprising:

a video image converting step of converting the input video image to a characteristics video image into which image characteristics have been extracted from the input video image;

a video image processing step of extracting areas which differ according to distances from the characteristics video image as video images classified by distance, on the basis of the distance from the vehicle, and composing a composite video image using the video images classified by distance, wherein the composite video image is a composite of said video images classified by distance;

a first object detecting step of performing first object detection processing of scanning the composite video image composed in the image processing step to determine a position on the composite video image of an object detected from the composite video image;

an object-candidate position detecting step of determining an object-candidate position where there is a possibility of existence of the object, in the video images classified by distance, on the basis of the position on the composite video image of the object detected in the first object detecting step;

a second object detecting step of performing, for the object-candidate position determined in the object-candidate position detecting step, second object detection processing for identifying corresponding object positions in the video images classified by distance; and

an object position identifying step of identifying an object position in the input video image on the basis of the object positions determined in the second object detecting step.

18. A non-transitory computer-readable medium storing an object detection program for detecting an object near a vehicle from an input video image, the input video image being a video image of surroundings of the vehicle shot from the vehicle, the program causing a computer to realize:

a video image converting function of converting the input video image to a characteristics video image into which image characteristics have been extracted from the input video image;

a video image processing function of extracting areas which differ according to distances from the characteristics video image as video images classified by distance, on the basis of the distance from the vehicle, and composing a composite video image using the video images classified by distance, wherein the composite video image is a composite of said video images classified by distance;

a first object detecting function of scanning the composite video image composed by the video image function to perform first object detection processing;

an object-candidate position detecting function of determining an object-candidate position where there is a possibility of existence of an object, in the video images classified by distance, on the basis of a position on the composite video image where an object has been detected by the first object detecting section;

a second object detecting function of performing second object detection processing for the object-candidate position determined by the object-candidate position detecting function to identify an object position in the video images classified by distance; and

an object position identifying function of identifying an object position in the input video image on the basis of the object positions determined by the second object detecting function.