Recognition device, recognition method, and recognition program

Abstract

A recognition device including: a detection unit that detects, from a video that includes a road, a non-road region, which is a region other than the road; and an assessment unit that assesses whether or not the non-road region includes video of an obstacle.

Claims

1. A detection unit that detects non-passage areas, which are areas that are not passages, from an image that includes a passage, The unit includes a determination unit that determines whether or not the non-passage area includes an image of an obstacle. The determination unit extracts feature quantities from the non-pathway region as vectors, and determines that the image does not contain an obstacle if the slope or length of the vectors are similar to each other, or if the vectors are pointing toward or in the opposite direction to the vanishing point. recognition device.

2. The determination unit extracts feature points from the non-passage area, and if the bias in the distribution of the feature points exceeds a predetermined value, it determines that the image does not contain any obstacles. The recognition device according to claim 1.

3. The determination unit estimates the dimensions, distance, or position of the non-passage area in real space from the viewpoint of the camera that captured the image and the coordinates of the non-passage area, and determines that the image does not contain an obstacle if the dimensions, distance, or position are outside a predetermined range. The recognition device according to claim 1.

4. The detection unit detects the passage region, which is the region of the convex hull that encompasses the entire passage. The determination unit determines that the image does not include an obstacle if the non-passage area is not included in the passage area and if the non-passage area is not located on the boundary line of the passage area. The recognition device according to claim 1.

5. The detection unit detects the height of the non-passage area, The determination unit determines that the image does not contain an obstacle if the height does not exceed a predetermined value. The recognition device according to claim 1.

6. From the video including the passageway, a non-passageway area, which is an area that is not a passageway, is detected. This includes determining whether the non-passage area includes an image of an obstacle, Determining whether or not the image of the aforementioned obstacle is included in extracting feature quantities as vectors from the non-pathway region, and determining that the image of the obstacle is not included if the slope or length of the vectors are similar to each other, or if the vectors are pointing toward or in the opposite direction to the vanishing point. A recognition method that causes a computer to perform a process.

7. From the video including the passageway, a non-passageway area, which is an area that is not a passageway, is detected. This includes determining whether the non-passage area includes an image of an obstacle, Determining whether or not the image of the aforementioned obstacle is included in extracting feature quantities as vectors from the non-pathway region, and determining that the image of the obstacle is not included if the slope or length of the vectors are similar to each other, or if the vectors are pointing toward or in the opposite direction to the vanishing point. A recognition program that causes a computer to perform a process.

Images