Image processing device

Abstract

To estimate a three-dimensional position of a subject with a small data transmission amount.SOLUTION: An image processing device includes a computation device. The computation device: detects an object in each of a plurality of images captured by a plurality of cameras; estimates a distance to the object in each of the plurality of images; uses the estimated distance to the object in each of the plurality of images and camera parameters of the plurality of cameras to determine, among feature points of the object, a feature point that can be acquired in any one of the plurality of images; derives a feature amount for the vicinity of the determined feature point in each of the plurality of images; compares the feature amounts of the plurality of images; and measures the distance to the object on the basis of results of comparing the feature amounts.SELECTED DRAWING: Figure 8

Claims

1. An image processing apparatus equipped with a computing device, The aforementioned computing device is Objects are detected in each of the multiple images captured by multiple cameras. For each of the aforementioned multiple images, a first distance is estimated, which is the distance from the camera that captured the image to the object. Using the first distance estimated for each of the multiple images and the camera parameters of the multiple cameras, the feature points of the object that can be obtained in any of the multiple images are determined. For each of the aforementioned multiple images, the feature quantities of the determined feature points are obtained. The feature quantities of the aforementioned multiple images are compared, Based on the results of matching the aforementioned feature quantities, a second distance, which is the distance from the image processing device to the object, is measured. Image processing device.

2. In the image processing apparatus according to claim 1, The aforementioned computing device is For each of the multiple images, a three-dimensional bounding box is determined that encloses the object. Of the multiple surfaces constituting the bounding box, the target surface that can be photographed by the multiple cameras is determined. To determine the feature quantities of the feature points included in the target surface, Image processing device.

3. In the image processing apparatus according to claim 1, The aforementioned computing device is For each of the multiple images, a three-dimensional bounding box is determined that encloses the object. For the multiple surfaces constituting the bounding box, the projection transformation between images captured by the multiple cameras is estimated. Based on the degree of deformation in the projection transformation of the aforementioned plurality of surfaces estimated above, the target surface from which to obtain the feature points is determined. To determine the feature quantities of the feature points included in the aforementioned target surface, Image processing device.

4. In the image processing apparatus according to claim 1, The calculation device determines the position from which to acquire the feature points of the object based on the object detection result and object shape information that associates the type of object with the degree of change in the object's shape over time. Image processing device.

5. In the image processing apparatus according to claim 1, The aforementioned computing device is A first computing resource to which some of the cameras among the aforementioned multiple cameras are connected, The system comprises a second computing resource to which a subset of cameras, different from some of the aforementioned plurality of cameras, are connected, Each of the first computing resource and the second computing resource is: The aforementioned object is detected, The first distance is estimated, Determine the aforementioned feature points, To determine the feature quantities of the aforementioned determined feature points, Image processing device.

Images