Object detection apparatus

• US 9,477,882 B2
• Filed: 03/13/2015
• Issued: 10/25/2016
• Est. Priority Date: 03/13/2014
• Status: Active Grant

• Current Assignee: Megachips Corporation
• Original Assignee: Megachips Corporation
• Inventors: Haraguchi, Yuki, Tani, Yujiro
• Primary Examiner(s): Abdi, Amara
• Application Number:
  US14/656,919
• Publication Number:
  US 20150262363A1
• Time in Patent Office: 592 Days
• US Class Current: 1/1
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 20/10 Terrestrial scenes scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

In an object detection apparatus (1), a window setting unit (12) extracts a window (52) from an edge image (22), to thereby generate an ordinary window image (23). When an edge strength of the ordinary window image (23) is not larger than a usage reference value, a used image determination unit (15) determines to use the ordinary window image (23) for calculation of an identification value (30) indicating a degree of presence of a pedestrian in the window (52). When the edge strength of the ordinary window image (23) is larger than the usage reference value, a pixel value correction unit (13) generates a corrected window image (24) having an edge strength smaller than that of the ordinary window image (23) from the ordinary window image (23). When the edge strength of the corrected window image (24) is not larger than the usage reference value, the used image determination unit (15) determines the corrected window image (24) to be used for calculation of the identification value (30).

Claims

1. An object detection apparatus for detecting an object to be detected from a frame image, comprising:

circuitry configured to;

set a window to be used for detection of the object to be detected in the frame image and acquire a first window image having an image of the window;

generate second window image by acquiring a value smaller than a pixel value of a pixel inside the first window image acquired by the circuitry when the pixel value is smaller than a correction reference value;

calculate a feature amount of the second window image generated by the circuitry;

determine whether or not to use the second window image for calculation of an identification value indicating whether or not the object to be detected is present in the window based on the feature amount calculated by the circuitry; and

calculate the identification value indicating a degree of presence of the object to be detected inside the second window image by using feature data of the object be detected when the circuitry determines to use the second window image.

2. The object detection apparatus according to claim 1, wherein the circuitry is further configured to:

compare the pixel value of the pixel inside the first window image with a reference value different from the correction reference value when the circuitry determines not to use the second window image, and generate a third window image by acquiring a value smaller than the pixel value when the pixel value is smaller than a difference reference value;

calculate a feature amount of the third window image generated by the circuitry;

determine whether or not to use the third window image based on the feature amount of the third window image, which is calculated by the circuitry; and

calculate an identification value indicating a degree of presence of the object to be detected inside the third window image when the circuitry determines to use the third window image.

3. The object detection apparatus according to claim 1, wherein the circuitry is further configured to:

calculate a feature amount of the first window image;

determine whether or not to use the first window image based on the feature amount of the first window image, which is calculated by the circuitry;

calculate an identification value indicating a degree of presence of the object to be detected inside the first window image when the circuitry determines to use the first window image; and

generate the second window image when the circuitry determines not to use the first window image.

4. The object detection apparatus according to claims 1, wherein

the frame image is an edge image,

the feature amount of the second window image is an edge strength, and

the circuitry is further configured to determine to use the second window image when the feature amount of the second window image is smaller than a usage reference value which is set in advance.

5. The object detection apparatus according to claim 1, wherein the circuitry is further configured to:

compare the pixel value of the pixel inside the first window image with a first reference value which is larger than the correction reference value, and generate a third window image by acquiring a pixel value smaller than the pixel value when the pixel value is smaller than the first reference value;

calculate a feature amount of the third window image; and

select a window image to be used for calculation of the identification value out of the second window image and the third window image based on the feature amount of the second window image and the feature amount of the third window image.

6. The object detection apparatus according to claim 5, wherein the circuitry is further configured to:

calculate a feature amount of the first window image; and

select a window image to be used for calculation of the identification value out of the first to third window images based on the respective feature amounts of the first to third window images.

7. The object detection apparatus according to claim 6, wherein the circuitry is further configured to:

calculate a first differential value indicating a difference between the feature amount of the first window image and the feature amount of the second window image and a second differential value indicating a difference between the feature amount of the second window image and the feature amount of the third window image,

wherein the circuitry selects a window image to be used for calculation of the identification value out of the first to third window images based on the first differential value and the second differential value.

8. The object detection apparatus according to claim 7, wherein

the circuitry is further configured to select the second window image when the first differential value is larger than the second differential value.

9. The object detection apparatus according to claim 5, wherein the circuitry is further configured to:

compare the pixel value of the pixel inside the first window image with a second reference value which is larger than the first reference value, and generate a fourth window image by acquiring a pixel value smaller than the pixel value when the pixel value is smaller than the second reference value;

calculate a feature amount of the fourth window image; and

select a window image to be used for calculation of the identification value out of the second to fourth window images based on the respective feature amounts of the second to fourth window images.

10. The object detection apparatus according to claim 1, wherein

the frame image is an edge image, and

the circuitry is further configured to calculate an edge strength of the second window image as the feature amount of the second window image.

11. The object detection apparatus according to claim 1, wherein the circuitry is further configured to:

compare the pixel value of the pixel inside the first window image with a zero reference value which is smaller than the correction reference value;

acquire a value obtained by replacing the pixel value with zero when the pixel value is smaller than the zero reference value; and

acquire a value obtained by replacing the pixel value with a value larger than zero when the pixel value is not smaller than the zero reference value and smaller than the correction reference value.

12. A non-transitory computer-readable recording medium storing an object detection program for causing a computer included in an object detection apparatus for detecting an object to be detected from a frame image to execute the steps of:

setting a window to be used for detection of the object to be detected in the frame image and acquiring a first window image having an image of the window;

generating a second window image by acquiring a value smaller than a pixel value of a pixel inside the first window image when the pixel value is smaller than a correction reference value;

calculating a feature amount of the second window image;

determining whether or not to use the second window image for calculation of an identification value indicating whether or not the object to be detected is present in the window based on the feature amount of the second window image; and

calculating the identification value indicating a degree of presence of the object to be detected inside the second window image by using feature data of the object to he detected when it is determined to use the second window image.