Apparatus for monitoring vicinity of a vehicle
View Patent ↗Category determination is performed on an object candidate extracted from a captured image after an exclusion process is performed in advance. In the exclusion process, a distance from a vehicle to the object candidate is calculated based on a size of the object candidate in a current image, a moved distance of the vehicle occurring in a time interval from a time a previous image was captured to a time the current image is captured is calculated, the size of the object candidate in the current image is decreased at a change ratio based on the distance to the object candidate and the moved distance, and a size of the object candidate in the previous image and a size of the object candidate decreased at the change ratio are compared. The object candidate with difference between them being determined to be larger than a predetermined value is removed.
1. An apparatus mounted on a vehicle for monitoring vicinity of the vehicle, comprising:
imaging means for capturing a time series of images of the vicinity of the vehicle;
extracting means for extracting an object candidate from the captured time series of images;
means for calculating distance (Z 0 or Z 1 ) from the vehicle to the object candidate at one point of time from the image captured at the one point of time under the assumption that the extracted object candidate has a size which is predetermined for an object of a desired category, and calculating distance (Z 0 or Z 1 ) from the vehicle to the object candidate at an other point of time on the basis of a velocity of the vehicle and the distance (Z 0 or Z 1 ) at the one point of time;
means for producing a virtual object which has a size (Ghv) determined by multiplying the size (Ghd 0 or Ghd 1 ) of the object candidate in the image captured at the one point of time by a ratio of the distance at the one point of time and the other point of time;
comparing means for generating a comparison of the size (Ghv) of the virtual object with the size (Ghd 1 or Ghd 0 ) of the object candidate in the image captured at the other point of time; and
means for determining that the extracted object candidate is an object of the desired category if the comparison substantially matches.
2. The apparatus of claim 1 , further comprising:
means for removing the object candidate whose size does not match the size of the virtual object from further processing.
3. The apparatus of claim 1 , wherein the means for determining determines that the comparison substantially matches if the comparison falls within a predetermined error range.
4. The apparatus of claim 1 , wherein the desired category of object is a human and the predetermined size is an average size of a human.
5. An apparatus mounted on a vehicle for monitoring vicinity of the vehicle, the apparatus having an image processing unit including a processor and a memory, and a imaging device for capturing a time series of images of the vicinity of the vehicle, the image processing unit being configured to:
extract an object candidate from the captured time series of images;
calculate distance (Z 0 or Z 1 ) from the vehicle to the object candidate at one point of time from the image captured at the one point of time under the assumption that the extracted object candidate has a size which is predetermined for an object of a desired category, and calculate distance (Z 0 or Z 1 ) from the vehicle to the object candidate at an other point of time on the basis of a velocity of the vehicle and the distance (Z 0 or Z 1 ) at the one point of time;
produce a virtual object which has a size (Ghv) determined by multiplying the size (Ghd 0 or Ghd 1 ) of the object candidate in the image captured at the one point of time by a ratio of the distance at the one point of time and the other point of time;
generate a comparison of the size (Ghv) of the virtual object with the size (Ghd 1 or Ghd 0 ) of the object candidate in the image captured at the other point of time; and
determine that the extracted object candidate is an object of the desired category if the comparison substantially matches.
6. The apparatus of claim 5 , wherein the image processing unit is further configured to remove the object candidate whose size does not match the size of the virtual object from further processing.
7. The apparatus of claim 5 , wherein the image processing unit determines that the comparison substantially matches if the comparison falls within a predetermined error range.
8. The apparatus of claim 5 , wherein the desired category of object is a human and the predetermined size is an average size of a human.
9. A method for monitoring vicinity of a vehicle, comprising:
capturing a time series of images of the vicinity of the vehicle;
extracting an object candidate from the captured time series of images;
calculating distance (Z 0 or Z 1 ) from the vehicle to the object candidate at one point of time from the image captured at the one point of time under the assumption that the extracted object candidate has a size which is predetermined for an object of a desired category, and calculating distance (Z 0 or Z 1 ) from the vehicle to the object candidate at an other point of time on the basis of a velocity of the vehicle and the distance (Z 0 or Z 1 ) at the one point of time;
producing a virtual object which has a size (Ghv) determined by multiplying the size (Ghd 0 or Ghd 1 ) of the object candidate in the image captured at the one point of time by a ratio of the distance at the one point of time and the other point of time;
generating a comparison of the size (Ghv) of the virtual object with the size (Ghd 1 or Ghd 0 ) of the object candidate in the image captured at the other point of time; and
determining that the extracted object candidate is an object of the desired category if the comparison substantially matches.
10. The method of claim 9 , further comprising:
removing the object candidate whose size does not match the size of the virtual object from further processing.