A Key Point is an ”interesting” point on the 2D photograph which the computer program determines to be an important feature. It is often a point of high contrast or interesting texture.
From SfM Field Methods Manual, UNAVCO (Indiana University Geological Field Station) 2016:
The software identifies features present in multiple photographs, called keypoint descriptors (hereafter referred to as key points) using an algorithm called SIFT (Scale-Invariant Feature Transform). Keypoints are used to calculate the relative locations of the cameras to create a sparse “low density” point cloud. A high-density point cloud is then generated based on the locations of the sparse points and the locations of the camera.
The Key point limit sets the standard establishing how finely Photoscan is supposed to sample each photo. The number indicates upper limit of the number of feature points on a photo PhotoScan will sample and will later be used in the alignment process. The larger the number the more the number of points PhotoScan will sample, potentially increasing accuracy. Using zero value allows PhotoScan to find as many key points as possible, but it may result in a large number of unreliable points. The recommended Agisoft starting point is 40,000, though for many images a setting of 50,000 is optimal. One investigator ran tests finding that values about 40,000 do little good, though it has been found that it is possible to rescue an image by using a higher amount.
Here's just a little extra for your dime:
A Tie Point is a Key Point that can be identified on two or more images that can be used to compute a 3D position.
Tie point limit: Tie points are those points found common to the photographs. The Tie point limit sets the standard for the upper limit in the number of tie points that will be used to tie one image to another. Tie point limit parameter optimizes performance and generally does not affect model quality. Too high or too low tie point limit value may cause some parts of the dense point cloud model to be missed. Using 0 value doesn't apply any tie point filtering so can take considerable time, though can resuscitate a problematic image. The number of tie points can be reduced after the alignment process with Tools/Tie Points - Thin Point Cloud. The sparse point cloud will be thinned, yet the alignment will remain unchanged. The recommended amount is 4,000, increasing if the object is small and there is a need for high accuracy. Some recommend using a Tie Point Limit of 0. Some have found that there is little savings in time with a broad range of values.
Topic: Key Points Meaning (Read 19194 times)
