Target object estimating apparatus

A target object estimating method comprising:

• capturing, by a stereo camera, an image of a region in front of a vehicle every time a predetermined time elapses;
• calculating, by a processor, on a basis of a reference image which is one of captured images used as a reference, the captured images being captured at a same computation timing by the stereo camera, parallax information including a parallax between corresponding pixels of the captured images for each of pixels in the reference image every time the predetermined time elapses;
• estimating, by the processor, whether each parallax point in the parallax information is a road surface parallax point or a parallax point other than a road surface parallax point every time the predetermined time elapses based on a deviation of magnitudes of the parallax points;
• acquiring, by the processor, vehicle state information and estimating a travelling expected route on which the vehicle is expected to travel based on the acquired vehicle state information;
• detecting, by the processor, a road surface region in the reference image based on the parallax information and the estimated travelling expected route every time the predetermined time elapses;
• determining, by the processor, whether or not each of the pixels in the reference image is a pixel corresponding to the road surface region every time the predetermined time elapses;
• calculating, by the processor, transition information including a change amount of a position from a position of a pixel in a previous reference image at a previous computational interval to a position of a pixel in the reference image, which is at a current computational interval, the pixel at the current computational interval corresponding to the pixel at the previous computational, for each of the pixels corresponding to a region other than the road surface region in the reference image every time said predetermined time elapses;
• and
• calculating, by the processor, estimated values including an estimated position and an estimated speed of a target object by calculating, based on the parallax information and the transition information, estimated values of the position and the speed in a three-dimensional space of each of the pixels corresponding to a region other than the road surface region in the reference image by applying a filter every time the predetermined time elapses.

A non-transitory computer-readable storage medium having computer-executable instructions stored thereon, wherein the computer-executable instructions, when executed by a processor, cause the processor to:

• capture by a stereo camera, an image of a region in front of a vehicle every time a predetermined time elapses;
• calculate on a basis of a reference image which is one of captured images used as a reference, the captured images being captured at a same computation timing by the stereo camera, parallax information including a parallax between corresponding pixels of the captured images for each of pixels in the reference image every time the predetermined time elapses;
• estimate whether each parallax point in the parallax information is a road surface parallax point or a parallax point other than a road surface parallax point every time the predetermined time elapses based on a deviation of magnitudes of the parallax points;
• acquire vehicle state information and estimating a travelling expected route on which the vehicle is expected to travel based on the acquired vehicle state information;
• detect a road surface region in the reference image based on the parallax information and the estimated travelling expected route every time the predetermined time elapses;
• determine whether or not each of the pixels in the reference image is a pixel corresponding to the road surface region every time the predetermined time elapses;
• calculate transition information including a change amount of a position from a position of a pixel in a previous reference image at a previous computational interval to a position of a pixel in the reference image, which is at a current computational interval, the pixel at the current computational interval corresponding to the pixel at the previous computational, for each of the pixels corresponding to a region other than the road surface region in the reference image every time said predetermined time elapses;
• and
• calculate estimated values including an estimated position and an estimated speed of a target object by calculating, based on the parallax information and the transition information, estimated values of the position and the speed in a three-dimensional space of each of the pixels corresponding to a region other than the road surface region in the reference image by applying a filter every time the predetermined time elapses.

A target object estimating apparatus mounted on a vehicle, the target object estimating apparatus comprising:

• a stereo camera configured to capture an image of a region in front of the vehicle every time a predetermined time elapses;
• and
• a processor configured to, for every computational interval of a plurality of computation intervals:
• obtain a left image and a right image captured by the stereo camera for the computational interval;
• calculate a parallax between corresponding pixels of the left image and the right image for each of a plurality of pixels of a reference image, which is one of the left image and the right image;
• generate a parallax image based on the calculated parallaxes for each of the pixels of the reference image;
• determine a group of the pixels composing the parallax image that are distributed in a range that rises in a longitudinal direction and partially converges in a lateral direction as a distance from the vehicle increases;
• classify the group of the pixels as a road surface region;
• acquire vehicle state information and estimate a travelling expected route on which the vehicle is expected to travel based on the acquired vehicle state information;
• determine whether or not each of the pixels in the reference image is a pixel corresponding to the road surface region every time the predetermined time elapses;
• calculate transition information including a change amount of a position from a position of a pixel in a previous reference image at a previous computational interval to a position of a pixel in the reference image, which is at a current computational interval, the pixel at the current computational interval corresponding to the pixel at the previous computational interval, for each of the pixels corresponding to a region other than the group of the pixels that corresponds to the road surface region in the reference image every time said predetermined time elapses;
• and
• calculate estimated values including an estimated position and an estimated speed of a target object by calculating, based on parallax information and the transition information, estimated values of the position and the speed in a three-dimensional space of each of the pixels corresponding to a region other than the road surface region in the reference image by applying a filter every time the predetermined time elapses.