| 摘要 |
By a disparity computation unit 21, first, with respect to a block in a bottom row of a first image, a corresponding block within a second image which has a same Y-coordinate as a Y-coordinate of the block and in which an object same as or similar to an object appearing in the block appears is searched for, and, next, in the case of searching for, with respect to a block which is upward in order from the searched block of the first image, a corresponding block within the second image in which an object same as or similar to that of the block appears, the corresponding block within the second image which has a same Y-coordinate as a Y-coordinate of the block of the first image and in which an object same as or similar to that of the block of the first image appears in an X-coordinate setting range with an X-coordinate of the corresponding block of the second image corresponding to the searched block which is immediately below the block of the first image as a center is searched for. Thus, provided is an obstacle detection device capable of simplifying processing for finding an obstacle, shortening a processing time and simplifying a hardware configuration. |
| 主权项 |
1. An obstacle detection device, comprising:
a stereo camera; disparity computation processing hardware that computes, based on a first image and a second image photographed by the stereo camera, disparity of an object appearing in the first and second images; and obstacle detection processing hardware that obtains a distance to the object based on the disparity of the object, which is computed by the disparity computation processing hardware, to detect whether or not the object is an obstacle, wherein lateral directions of the first and second images are set as X-coordinates and vertical directions of the first and second images are set as Y-coordinates, the first image is divided into a plurality of blocks that are arrayed in a matrix shape, and the disparity computation processing hardware:
searches for, with respect to a first block of the plurality of blocks in a bottom row of the first image, a first corresponding block within the second image which has a same Y-coordinate as a Y-coordinate as the first block and in which the object appearing in the first block appears,searches for, with respect to a second block of the plurality of blocks that is upward from the first block, a second corresponding block within the second image in which the object appears, and searches for the second corresponding bock within the second image which has a same Y-coordinate as a Y-coordinate of the second block in which the object appears in an X-coordinate setting range with an X-coordinate of the second corresponding block of the second image corresponding to the first block which is immediately below the second block of the first image as a center,computes disparity of the object as it appears in the first and second images from a difference between an X-coordinate of the first block and the X-coordinate of the first corresponding block,computes a distance to a road surface in the first block based on a vertical view angle, a depression angle, and a camera height of the stereo camera,computes disparity between the road surface in the first block and the road surface in the second image based on the distance to the road surface, andsearches for a corresponding block within the second image which has a same Y-coordinate as a Y-coordinate of the first block and in which the object appears in the X-coordinate setting range with an X-coordinate range obtained by subtracting the display from an X-coordinate of the first block as a center. |
