Apparatus and method for detecting upper body

• US 8,705,814 B2
• Filed: 12/21/2011
• Issued: 04/22/2014
• Est. Priority Date: 12/21/2010
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: Yun, Woo Han, Kim, Do Hyung, Lee, Jae Yeon, Ban, Kyu Dae, Lee, Dae Ha, Han, Mun Sung, Yoon, Ho Sub, Chi, Su Young, Chung, Yun Koo, Sohn, Joo Chan, Kim, Hye Jin, Yoon, Young Woo, Kim, Jae Hong, Cho, Jae Il
• Primary Examiner(s): Le, Vu
• Assistant Examiner(s): WOLDEMARIAM, AKLILU K
• Application Number:
  US13/332,879
• Publication Number:
  US 20120155719A1
• Time in Patent Office: 853 Days
• Field of Search: 382/165, 382/128, 382/118, 382/103
• US Class Current: 382/118
• CPC Class Codes: G06V 40/10 Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands

Abstract

Disclosed is a method of detecting an upper body. The method includes detecting an omega candidate area including a shape formed of a face and a shoulder line of a human from a target image, cutting the target image into the upper body candidate area including the omega candidate area, detecting a human face from the upper body candidate area, and judging whether the upper body of the human is included in the target image according to the result of detecting the human face.

Claims

1. An apparatus for detecting an upper body, comprising:

an omega field detecting unit configured to detect an omega field including a shape formed of a face and a shoulder line of a human from a target image, the omega field detecting unit including;

a resizing unit configured to reduce a width or a length of the target image by a predetermined percentage;

a first feature deducing unit configured to deduce a weight of coordinates of the reduced target image and deduce a first feature on the basis of the deduced weight; and

an omega field judging unit configured to judge whether the omega field is included in the target image by comparing the first feature with a reference value;

a face detecting unit configured to detect whether the human face is included in the omega field;

a side/rear verifying unit configured to verify whether a side or a rear of a human exists in the omega field if the human face is not detected by the face detecting unit; and

an upper body judging unit configured to judge whether the upper body of a human is included in the target image on the basis of a result of detecting the human face and configured to judge whether the upper body of a human is included in the target image on the basis of a result of verifying whether a side or a rear of a human exists in the omega field.

2. The apparatus of claim 1, wherein the omega field detecting unit further includes:

a first memory configured to store a three-dimensional first lookup table that is trained based.

3. The apparatus of claim 2, wherein the first lookup table is trained and deduced based on the adaboost algorithm using an omega shape example image as a positive set and the other images than the omega shape example image as a negative set.

4. The apparatus of claim 1, wherein the side/rear verifying unit includes:

a second memory configured to store a three-dimensional second lookup table trained based on the adaboost algorithm;

a second feature deducing unit configured to deduce a weight of coordinates of the omega field based on the second lookup table and deduce a second feature based on the deduced weight; and

a side/rear judging unit configured to judge whether the side or the rear of a human is included in the omega field by comparing the second feature with a reference value.

5. The apparatus of claim 4, wherein the second lookup table is trained and deduced based on the adaboost algorithm using a human side/rear example image as a positive set and the other images than the human side/rear example image as a negative set.

6. A method for detecting an upper body, comprising:

detecting an omega candidate area including a shape formed of a face and a shoulder line of a human from a target image;

cutting the target image into the upper body candidate area including the omega candidate area;

detecting whether a human face appears in the upper body candidate area;

detecting whether a side or a rear of a human appears in the upper body candidate area if the human face is not detected in the upper body candidate area, and

judging whether the upper body of the human is included in the target image according to a result of detecting the human face and according to a result of detecting wherein the omega candidate area is detected by;

reducing a width or a length of the target image by a predetermined percentage,

deducing a weight of coordinates of the reduced target image and deducing a first feature on the basis of the deduced weight, and

judging whether the omega field is included in the target image by comparing the first feature with a reference value.

7. The method of claim 6, wherein the judging includes:

if the human face is detected in the detecting of the human face, judging that the human upper body is included in the target image.

8. The method of claim 6, wherein the detecting includes:

determining whether a width or a length of the target image is larger than a predetermined threshold TH;

reducing the size of the target image by a predetermined percentage if the width or the length of the target image is larger than the predetermined threshold;

pre-processing the resized target image;

deducing a first feature from the pre-processed target image; and

judging if the omega candidate area is included in the target image based on the deduced first feature.

9. The method of claim 6, wherein the detecting of the side or the rear of a human includes:

pre-processing the upper body candidate area, the pre-processing including any one of an edge detecting process and an MCT process;

deducing a second feature from the pre-processed upper body candidate area; and

judging whether the side or the rear of a human exists in the upper body candidate area by comparing the second feature with a predetermined reference value.