Object recognition apparatus and method

An object recognition apparatus mounted on a vehicle, said apparatus comprising:

• an on-vehicle camera acquiring an image around the vehicle;
• a position detection device including sensors and road map data, said position detection device detecting first support information including a bearing, a position, and a mileage of the vehicle;
• means for estimating an image cut off range in the image acquired by said on-vehicle camera on the basis of the first support information supplied from the position detection device, said image cut off range is assumed to include an object to be recognized;
• and
• means for recognizing the object in the image cut off range,
• wherein said means for estimating an image cut off range comprises:
• means for estimating, on the basis of the position of the object in the image at a particular time, the position of the object in the image at the next time, the interval between the particular time and the next time being determined in advance;
• and
• means for determining the image cut off range on the basis of the estimated position, said means for estimating an image cut off range providing the image cut off range with an error range that allows a certain amount of error in determining the image cut off range,
• wherein said position detection device has information associated with the speed of the vehicle, and said means for estimating an image cut off range determines the error range in accordance with the speed of the vehicle.

An object recognition apparatus mounted on a vehicle, said apparatus comprising:

• an on-vehicle camera acquiring an image around the vehicle;
• a position detection device including sensors and road map data, said position detection device detecting first support information including a bearing, a position, and a mileage of the vehicle;
• means for estimating an image cut off range in the image acquired by said on-vehicle camera on the basis of the first support information supplied from the position detection device, said image cut off range is assumed to include an object to be recognized;
• and
• means for recognizing the object in the image cut off range,
• wherein said means for estimating an image cut off range comprises:
• means for estimating, on the basis of the position of the object in the image at a particular time, the position of the object in the image at the next time, the interval between the particular time and the next time being determined in advance;
• and
• means for determining the image cut off range on the basis of the estimated position, said means for estimating an image cut off range providing the image cut off range with an error range that allows a certain amount of error in determining the image cut off range, wherein said position detection device provides said means for estimating an image cut off range with information associated with the bearing of the vehicle, and said means for estimating an image cut off range determines the error range in accordance with an amount of change in the bearing of the vehicle, and wherein said means for estimating an image cut off range determines the error range such that the error range is substantially proportional to the change in the bearing of the vehicle when the change of the bearing of the vehicle per unit time or per unit mileage is greater than a predetermined value.

An object recognition apparatus mounted on a vehicle, said apparatus comprising:

• an on-vehicle camera acquiring an image around the vehicle;
• a position detection device including sensors and road map data, said position detection device detecting first support information including a bearing, a position, and a mileage of the vehicle;
• means for estimating an image cut off range in the image acquired by said on-vehicle camera on the basis of the first support information supplied from the position detection device, said image cut off range is assumed to include an object to be recognized;
• means for recognizing the object in the image cut off range;
• camera attitude parameter generating means for generating camera attitude parameters representing the attitude of the on-vehicle camera with respect to the road;
• and
• reliability obtaining means for obtaining reliability of the camera attitude parameters generated by said camera attitude parameter generating means,
• wherein said means for estimating the position of the object estimates the position of the object in the image using the camera attitude parameters, and means for estimating the image cut off range estimates the error range on the basis of the reliability of the camera attitude parameters, and wherein said means for estimating an image cut off range comprises:
• means for estimating, on the basis of the position of the object in the image at a particular time, the position of the object in the image at the next time, the interval between the particular time and the next time being determined in advance;
• and
• means for determining the image cut off range on the basis of the estimated position, said means for estimating an image cut off range providing the image cut off range with an error range that allows a certain amount of error in determining the image cut of range.

A tracking recognition apparatus performing a tracking recognition of an object on and around a road, on which a vehicle is traveling, on the basis of an image acquired by an on-vehicle camera which picks up an image ahead of or behind the vehicle, said tracking recognition apparatus comprising:

• image converting means for converting the image acquired by said on-vehicle camera into a digital image signal at every processing period;
• means for generating contours of one or more objects in the image on the basis of the digital image signal supplied from the image converting means;
• means for determining coordinates of one or more specified points in the image, said specified points being selected in advance for each object to be recognized so that the position of each object is substantially determined by the specified points;
• registration means for registering the specified points to a memory with characteristics of a neighboring image of the specified points;
• means for generating camera attitude parameters of the on-vehicle camera, said camera attitude parameters representing the attitude of the on-vehicle camera with respect to the road;
• spatial coordinate converting means for converting the coordinates of the specified points registered in said registration means into spatial positions of the specified points by using the camera attitude parameters of the on-vehicle camera;
• a position detection device detecting the position of the vehicle, and producing mileage data and bearing data of the vehicle;
• first estimating means for estimating spatial positions of the specified points at the next processing period on the basis of the mileage data, the bearing data of the vehicle, and the spatial positions of the specified points obtained by the spatial coordinate converting means;
• second estimating means for estimating the coordinates of the specified points at the next processing period in the image associated with the spatial positions;
• means for determining an error range around the coordinates of the specified points estimated by said second estimating means when an image of the next processing period is outputted from the image converting means;
• and
• decision means for deciding whether the same specified points are present in the error range in the image of the next processing period on the basis of the characteristics of the neighboring image of the specified points,
• wherein if the decision means decides that the same specified points are present, the object is tracked by obtaining the spatial positions of the specified points, whereas if the decision means decides that the same specified points are absent, the registration associated with the object is canceled.

An object recognition method for use with a vehicle comprising the steps of:

• acquiring an image around the vehicle by an on-vehicle camera;
• detecting, by using sensors and road map data, first supporting information including a bearing, a position and a milage of the vehicle;
• estimating an image cut off range in said image acquired by said on-vehicle camera, on the basis of the first support information supplied from the position detection device, said image cut off range is assumed to include an object to be recognized;
• recognizing the object in the image cut off range;
• and
• calculating a plurality of camera attitude parameters which indicate attitude of said on-vehicle camera with respect to the road.

A tracking recognition method performing a tracking recognition of an object on and around a road, on which a vehicle is traveling, on the basis of an image acquired by an on-vehicle camera which picks up an image ahead of or behind the vehicle, said tracking recognition method comprising the steps of:

• converting the image acquired by said on-vehicle camera into a digital image signal at every processing period;
• generating contours of one or more objects in the image on the basis of the digital image signal;
• determining coordinates of one or more specified points in the image, said specified points being selected in advance for each object to be recognized so that the position of each object is substantially determined by the specified points;
• registering the specified points to a memory with characteristics of a neighboring image of the specified points;
• generating camera attitude parameters of the on-vehicle camera, said camera attitude parameters representing the attitude of the on-vehicle camera with respect to the road;
• converting the coordinates of the specified points registered into spatial positions of the specified points by using the camera attitude parameters of the on-vehicle camera;
• detecting the position of the vehicle, and producing mileage data and bearing data of the vehicle;
• estimating spatial positions of the specified points at the next processing period on the basis of the mileage data, the bearing data of the vehicle, and the spatial positions of the specified points;
• estimating the coordinates of the specified points at the next processing period in the image associated with the spatial positions;
• determining an error range around the coordinates of the specified points when an image of the next processing period is outputted;
• and
• deciding whether the same specified points are present in the error range in the image of the next processing period on the basis of the characteristics of the neighboring image of the specified points,
• wherein if the same specified points are present, the object is tracked by obtaining the spatial positions of the specified points, whereas if the same specified points are absent, the registration associated with the object is canceled.

An object recognition apparatus mounted on a vehicle comprising:

• an on-vehicle camera acquiring an image around the vehicle;
• a position detection device including sensors and road map data, said position detection device detecting first support information including a bearing, a position, and a mileage of the vehicle;
• means for estimating an image cut off range in the image acquired by said on-vehicle camera on the basis of the first support information supplied from the position detection device, said image cut off range is assumed to include an object to be recognized;
• and
• means for recognizing the object in the image cut off range, wherein said position detection device has information associated with attributes of the road on which the vehicle is running, and said means for recognizing the object determines the type of the object which is limited in accordance with the attributes of the road supplied from the position detection device.