Recognition device, recognition method, and recognition program
A recognition device that detects a boundary of a roadway by use of time-series images captured by a camera mounted on a vehicle, the recognition device including: an acquisition unit that acquires the time-series images; and a detection unit that classifies areas in the acquired time-series images into a roadway, which is an area having a small change amount, and a portion other than the roadway, which is an area having a large change amount, according to magnitude of the change amounts, and detects a boundary between the area having the small change amount and the area having the large change amount as a boundary of the roadway.
1 . A recognition device that detects a boundary of a roadway by use of time-series images captured by a camera mounted on a vehicle, the recognition device comprising:
a memory, and
at least one processor coupled to the memory, the at least one processor being configured to:
acquire the time-series images;
generate a plurality of images in which changes between the time-series images are emphasized by using the acquired time-series images;
calculate a difference image by superimposing the plurality of images in which the changes are emphasized;
calculate change amounts of areas based on changes in pixels of the difference image; and
clarify areas in the acquired time-series images into a roadway, which is an area having a small change amount, and a portion other than the roadway, which is an area having a large change amount, according to magnitude of the change amounts, and detect a boundary between the area having the small change amount and the area having the large change amount as a boundary of the roadway.
2 . The recognition device according to claim 1 , wherein the processor is configured to further perform correction to remove an area in which the vehicle itself is captured is not detected from the time-series images.
3 . The recognition device according to claim 1 , wherein the processor is configured to
acquire a speed of the vehicle, and
use only an image captured when the speed has a value equal to or larger than a predetermined threshold value.
4 . The recognition device according to claim 1 , wherein the processor is configured to correct one of binarized areas to another of the binarized areas in a case where a width of one of binarized areas of the difference image is narrower than a predetermined length.
5 . The recognition device according to claim 1 , wherein the processor is configured to acquire road information corresponding to the number of lanes of the roadway on which the vehicle is traveling, and specify a traveling lane of the vehicle and a peripheral traveling lane as the area having the small change amount on a basis of the road information and a division line extracted in the time-series images.
6 . A recognition method for causing a computer that detects a boundary of a roadway by use of time-series images captured by a camera mounted on a vehicle to execute processing comprising:
acquiring the time-series images;
generating a plurality of images in which changes between the time-series images are emphasized by using the acquired time-series images;
calculating a difference image by superimposing the plurality of images in which the changes are emphasized;
calculating change amounts of areas based on changes in pixels of the difference image; and
classifying areas in the acquired time-series images into a roadway, which is an area having a small change amount, and a portion other than the roadway, which is an area having a large change amount, according to magnitude of the change amounts, and detecting a boundary between the area having the small change amount and the area having the large change amount as a boundary of the roadway.
7 . A non-transitory, computer-readable storage medium storing a recognition program for causing a computer that detects a boundary of a roadway by use of time-series images captured by a camera mounted on a vehicle to execute processing comprising:
acquiring the time-series images;
generating a plurality of images in which changes between the time-series images are emphasized by using the acquired time-series images;
calculating a difference image by superimposing the plurality of images in which the changes are emphasized;
calculating change amounts of areas based on changes in pixels of the difference image; and
classifying areas in the acquired time-series images into a roadway, which is an area having a small change amount, and a portion other than the roadway, which is an area having a large change amount, according to magnitude of the change amounts, and detecting a boundary between the area having the small change amount and the area having the large change amount as a boundary of the roadway.