Calibrations for a Vision Based System
AssigneePrecision Planting LLC
InventorMichael Strnad
Life of this patent
Estimating the term…
Abstract
A computer implemented method for calibration of a camera comprises capturing, with the camera that is disposed on an implement, a sequence of images while the implement travels across a terrain, comparing a first image from an image sensor of the camera at a first time to a second image from the image sensor at a second time, determining matching points corresponding to features in common in the first image and in the second image, and determining at least one of height, pitch, roll, and yaw for the camera based on the first image, the second image, the matching points corresponding to features in common in the first image and in the second image, and a ground speed of the implement while capturing the first image at the first time and the second image at the second time.
Claims20
A computer implemented method for calibration of a camera comprising:
- capturing, with the camera that is disposed on an implement, a sequence of images while the implement travels across terrain;
- comparing a first image from an image sensor of the camera at a first time to a second image from the image sensor at a second time;
- determining matching points corresponding to features in common in the first image and in the second image;
- and
- determining at least one of height, pitch, roll, and yaw for the camera based on the first image, the second image, the matching points corresponding to features in common in the first image and in the second image, and a ground speed of the implement while capturing the first image at the first time and the second image at the second time.
A system comprising:
- an agricultural implement;
- a camera disposed on the agricultural implement, the camera is configured to capture a sequence of images while the agricultural implement travels through an agricultural field;
- and
- a processor that is configured to compare a first image from an image sensor of the camera at a first time to a second image from the image sensor at a second time, determine matching points corresponding to features in common in the first image and in the second image, and determine at least one of height, pitch, roll, and yaw for the camera based on the first image, the second image, the matching points corresponding to features in common in the first image and in the second image, and a ground speed of the agricultural implement while capturing the first image at the first time and the second image at the second time.
A computer implemented method for aligning a second image sensor with a first image sensor of a camera, comprising:
- using a calculated height, pitch, roll, and yaw based on captured images in an agricultural field for the camera that is disposed on an implement to calculate a real world projection matrix for the first image sensor to allow features, image points, or pixels from an image space to be projected into real world ground projected coordinates;
- determining a value for each corner point for the real world ground projected coordinates from the first image sensor;
- calculating a nominal disparity based on each of the corner points for the real world ground projected coordinates and corner points in image space of the first image sensor;
- warping a first image from the first image sensor by the nominal disparity for each of those corner points;
- and
- determining a registration matrix to align a second raw image from the second image sensor with the disparity warped first image.
Dependencies are inferred from the claim text.