Three-dimensional object detection device and three-dimensional object detection method

摘要

A three-dimensional object detection device has an image capturing unit, a first object detection unit, a light source detection unit, a second object detection unit, a degree-of-clouding calculation unit and a controller. The image capturing unit captures images rearward of a vehicle. The first object detection unit detects a presence of a three-dimensional object from the captured images. The second object detection unit detects a presence of a three-dimensional object from a detection result of the light source detection unit. The degree-of-clouding calculation unit calculates a degree of lens clouding from the captured images. The controller assess the object as an adjacent vehicle in an adjacent lane from the degree of lens clouding when the degree of lens clouding is equal to or greater than a predetermined value and from the presence of the object in the captured images when the degree of lens clouding is less than the value.

主权项

1. A three-dimensional object detection device comprising:
an image capturing unit having a lens for forming captured images of an area rearward of a host vehicle; and a computer operatively arranged to receive input of the captured images from the image capturing unit, the computer being programmed to
perform viewpoint conversion of the captured images captured by the image capturing unit to bird's-eye view images,detect a presence of a three-dimensional object based on the captured images captured by the image capturing unit, the presence of the three dimensional object being detected based on differential waveform information that is generated by
aligning positions of two of the bird's-eye view images obtained at two different times,calculating a pixel value difference with respect to each pixel of the aligned bird's-eye view images,generating a differential image based on the calculated pixel value differences,defining at least one line in the differential image, the at least one line being oriented along a direction in which the three-dimensional object appears collapsed in the bird's-eye view images,identifying difference pixels positioned on the at least one line that have a pixel value difference equal to or larger than a predetermined pixel value difference,counting a number of the difference pixels identified along each of the at least one line, andgenerating the differential waveform information as a frequency distribution based on the number of the difference pixels counted in each of the at least one line;detect a light source which is present behind the host vehicle based on the captured images captured by the image capturing unit,detect a presence of a three-dimensional object based on a detection result of detecting the light source,calculate a degree to which the lens is clouded as a degree of lens clouding based on the captured images, the degree of lens clouding being calculated based on a number of peaks having a value equal to or greater than a predetermined threshold value among peaks in the differential waveform information generated within a predetermined time interval, the degree of lens clouding indicating a degree to which a foreign matter is adhered to the lens,assess whether the three-dimensional object is an adjacent vehicle present in an adjacent lane adjacent to a lane of travel of the host vehicle based on at least a detection result of detecting the presence of the three-dimensional object based on the light source when the degree of lens clouding is equal to or greater than a predetermined determination value, andassess whether the three-dimensional object is the adjacent vehicle based on at least a detection result of detecting the presence of the three-dimensional object based on the captured images when the degree of lens clouding is less than the determination value.