imagePoints

imagePoints is a term used in computer vision to denote a set of two-dimensional points in an image that correspond to a set of known three-dimensional points from a scene or object. In calibration and 3D reconstruction workflows, objectPoints are the 3D coordinates in a chosen coordinate system, and imagePoints are the projected 2D coordinates in the image plane obtained by projecting objectPoints using the camera intrinsics and extrinsics. In OpenCV and similar libraries, imagePoints are typically stored as a vector of 2D points (for example, std::vector<cv::Point2f> or a Mat with size N x 1 x 2). They are used as inputs or outputs for functions such as calibrateCamera, stereoCalibrate, and solvePnP. For example, during camera calibration, one provides a set of objectPoints and the corresponding imagePoints for multiple views; the function estimates the camera matrix and distortion coefficients that map between objectPoints and imagePoints.

imagePoints are usually measured in pixel coordinates (u, v) with the origin at the image's top-left corner, and the units depend on the image resolution. The accuracy of imagePoints affects calibration quality and downstream 3D estimation; points should be well distributed across the image and cover the scene geometry. In practice, imagePoints are often obtained by detecting and matching features or pattern points (such as chessboard corners) across images, and then organizing the correspondences to align with the corresponding objectPoints. Proper handling of distortions and consistent coordinate conventions are important for reliable results.