Technique for classifying objects within an image

• US 6,556,708 B1
• Filed: 02/06/1998
• Issued: 04/29/2003
• Est. Priority Date: 02/06/1998
• Status: Expired due to Fees

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Compaq Computer Corporation (HP Inc.)
• Inventors: Avery, Brian Lyndall, Christian, Andrew Dean
• Primary Examiner(s): Wu, Jingge
• Application Number:
  US09/020,203
• Time in Patent Office: 1,908 Days
• Field of Search: 382/103, 382/104, 382/107, 382/165, 382/283, 382/105, 382/118, 382/115, 382/203, 382/209, 382/213, 382/224, 348/152, 348/155, 348/156, 348/169, 348/139, 348/143, 348/148-151, 348/153, 348/154, 348/135-138
• US Class Current: 382/165
• CPC Class Codes: G06V 40/10 Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands

Abstract

A technique for classifying objects within an image is disclosed. In one embodiment, the technique is realized by identifying a portion of an image, and then filtering the portion of the image based upon an object characteristic and a reference within the image. The image can be a representation of a plurality of pixels, wherein at least some of the plurality of pixels are enabled to represent the identified portion of the image. The reference can be one of a reference plane or a terrain within the image.

Claims

1. A method of image management comprising the steps of:

storing a plurality of source images;

dividing the plurality of source images into a first subset of the plurality of source images and a second subset of the plurality of source images;

selecting the first subset of the plurality of source images;

forming a background image using the selected first subset of the plurality of source images;

comparing a first source image included in the second subset of the plurality of source images with the formed background image, wherein the first source image is not included with the first subset of the plurality of images;

responsive to comparing the first source image with the formed background image, identifying object images included in the first source image; and

applying a differencing algorithm further comprises the steps of;

computing a Y-mask image;

computing a UV-mask image;

comparing the Y-mask image and the UV-mask image; and

computing a YUV mask image responsive to the step of comparing the Y-mask image and the UV-mask image;

using a known position and orientation of a video camera that has captured the plurality of source images to determine first object characteristics of the object images identified;

determining second object characteristics, the characteristics including size, shape and ratio of width to height;

applying a filter to determine if at least one of the first object characteristics of the identified object images corresponds with one second object characteristic;

selecting an identified object when the first object characteristic of the identified object image corresponds with the second object characteristic;

tracking the identified object through a second source image wherein the second source image is included in the second subset of the plurality of source images; and

updating the background image by including therein the first and the second source images less the identified object.

2. The method of claim 1, wherein the first subset of the plurality of source images includes only one source image.

3. The method of claim 1, wherein the step of forming a background image comprises the step of:

averaging together each of the individual source images included in the first subset of the plurality of images, thereby forming a time-averaged background image.

4. The method of claim 1, wherein the step of comparing the first source image with the formed background image comprises the step of:

applying a differencing algorithm to the first source image and the formed background image.

5. The method of claim 4, wherein the first source image includes Y, U, and V components and the formed background image includes Y, U, and V components and further wherein the step of applying the differencing algorithm comprises the steps of:

extracting the Y, U, and V components from the first source image;

extracting the Y, U, and V components from the formed background image;

determining the difference between the Y, U, and V components of the first source image and the Y, U, and V components of the formed background image; and

forming a difference image responsive to determining the difference between the Y, U, and V components of the first source image and the Y, U, and V components of the formed background image.

6. The method of claim 1, further comprising the step of:

computing an object location for the identified object image;

wherein the identified object image is selectable when the computed object location is within a threshold distance.

7. The method of claim 1, the method further comprising the steps of:

tracking the identified object image through a third source image, the third source image included in the second subset of the plurality of source images; and

computing a velocity of movement for the identified object image between the second source image and the third source image.

8. The method of claim 7, wherein the object characteristic includes velocity of movement and wherein the identified object is selectable when the velocity of movement is within a threshold range.

9. An apparatus for image management comprising:

at least one processor;

a storage device connected to the at least one processor, the storage device containing a plurality of instructions readable by the at least one processor to thereby cause the at least one processor to operate to;

store a plurality of source images;

divide the plurality of source images into a first subset of the plurality of source images and a second subset of the plurality of source images;

select the first subset of the plurality of source images;

form a background image using the selected first subset of the plurality of source images;

compare a first source image included in the second subset of the plurality of source images with the formed background image, wherein the first source image is not included with the first subset of the plurality of images; and

the instructions contained on the storage device are configured to cause the processor to;

compute a Y-mask image;

compute a UV-mask image;

compare the Y-mask image and the UV-mask image; and

compute a YUV mask image responsive to the step of comparing the Y-mask image and the UV-mask image;

responsive to the comparison of the first source image with the formed background image, identify object images included in the first source image;

a video camera to capture the plurality of source images that are stored, the video camera having known position and orientation that are used to determine first object characteristics of the object images identified;

determine second object characteristics, the characteristics including size, shape and ratio of width to height;

the processor then operates to apply a filter to determine if at least one of the first object characteristics of the identified object image corresponds with a second object characteristic;

select an identified object image when the first object characteristic of the identified object image corresponds with the second object characteristic;

track the identified object image through a second source image wherein the second source image is included in the second subset of the plurality of source images, and update the background by including therein the first and the second source images less the identified object.

10. The apparatus of claim 9, wherein the first subset of the plurality of source images includes only one source image.

11. The apparatus of claim 9, wherein the instructions contained on the storage device are configured to cause the processor to:

average together each of the individual source images included in the first subset of the plurality of images, thereby forming a time-averaged background image.

12. The apparatus of claim 9, wherein the instructions contained on the storage device are configured to cause the processor to:

apply a differencing algorithm to the first source image and the formed background image.

13. The apparatus of claim 12, wherein the first source image includes Y, U, and V components and the formed background image includes Y, U, and V components and wherein the instructions contained on the storage device are configured to cause the processor to:

extract the Y, U, and V components from the first source image;

extract the Y, U, and V components from the formed background image;

determine the difference between the Y, U, and V components of the first source image and the Y, U, and V components of the formed background image; and

form a difference image responsive to determining the difference between the Y, U, and V components of the first source image and the Y, U, and V components of the formed background image.

14. The apparatus of claim 9, wherein the instructions contained on the storage device are configured to cause the processor to:

compute an object location for the identified object image;

wherein the identified object image is selectable when the computed object location is within a threshold distance.

15. The apparatus of claim 9, wherein the instructions contained on the storage device are configured to cause the processor to:

track the identified object image through a third source image, the third source image included in the second subset of the plurality of source image; and

compute a velocity of movement for the identified object image between the second source image and the third source image.

16. The apparatus of claim 15, wherein the object characteristic includes velocity of movement and wherein the identified object is selectable when the velocity of movement is within a threshold range.