Add a soft constraint to keep the cameras near the original position. Add a combined constraint for all reprojection errors in given camera.

Add the constraint based on DEM

Add residual block for the error using reference xyz.

Add the frame camera model reprojection error to the cost function

Add the GCP constraint

Add a ground constraint (GCP or height from DEM)

Add hard camera constraints. Be generous with the uncertainty.

Add one JSON frame bounding box to the KML object

Add the linescan model reprojection error to the cost function

Pixel reprojection error. Note: cam_residual_counts and num_pixels_per_cam serve different purposes.

Add reprojection errors. Collect data that will be used to add camera constraints that scale with the number of reprojection errors and GSD.

Add error source for projecting a 3D point into the camera.

Add the camera position constraints for the ref cams

Try to make each mesh intersection agree with corresponding depth measurement, if it exists

Ensure that the depth points agree with triangulated points

Reprojection error with ls ref sensor and frame curr sensor

Reprojection error with ls ref sensor and frame curr sensor

Reprojection error with ref ls ref sensor and curr ls sensor TODO(oalexan1): Find ways of integrating with the frame

Add the ls or frame camera model reprojection error to the cost function

Add roll / yaw constraints. For linescan, use the whole set of samples for given camera model. For frame cameras, use the trajectory of all cameras in

Add a constraint that the curvature of the sequence of poses does not become a lot more than the initial one.

Add camera constraints that are proportional to the number of reprojection errors. This requires going through some of the same motions as in addRepro

* @brief Common-case helper set mesh vertex colors. Creates a vertex element, if necessary. * * @param colors A vector of vertex colors (unsigne

* @brief Common-case helper set mesh vertex positons. Creates vertex element, if necessary. * * @param vertexPositions A vector of vertex positi

Add a camera frustum wireframe to a mesh.

Expands the BBox in all directions by the given size

Given a list of strings, protect each one having spaces with quotes, if not present.

Add to the outliers based on the large residuals

Adjust the weights at the boundary of the max-lit region so they don't decay to 0, as there we need all the image data we get, and there's no concern

Adjust the orbit end point and set the number of cameras given the frame rate This is a bit tricky because need to sample finely the orbit

The images from the bag may need to be resized to be the same size as in the calibration file. Sometimes the full-res images can be so blurry that int

Change the rig to have desired offsets in the sensor plane and ground plane. These will determine the orientations. The offsets are in meters.

Extract a affine transform to an array of length NUM_AFFINE_PARAMS

Apply alignment transform to ip. Not to be used with mapprojected images.

The name of the prefix (sans folder) for the generated bundle-adjusted and pc_aligned camera files.

Create interest points from valid D_sub values and make them full scale (while still having potentially a global alignment applied to them).

Return true if all the required source data has been downloaded

Convert azimuth and elevation in degrees to Sun position in meters

Append a tile to a given set of polygons

Apply the camera2world transforms and camera centers to the cameras

Create cameras with the pc_align transform applied to them, # so that we can later generate ortho images.

This function takes advantage of the fact that when it is called the cam_ptrs have the same information as is in param_storage. This applies the trans

This function takes advantage of the fact that when it is called the cam_ptrs have the same information as is in param_storage.

Apply a given transform to a LAS file and save it.

Given a transform with origin at the planet center, like output by pc_align, read the adjustments from param storage, apply this transform on top of t

Apply transforms to ip

The approximate inverse based on tabulated values. It can return nan.

Use existing interest points to compute a search range, when --seed-mode is 1.

Attempt to automatically create the overlap list file estimated footprints for each of the input images. - Currently this only supports cameras with W

Compute a normal vector based on the azimuth and elevation angles

Find the bounding box of the camera centers

Find the best fitting plane to a set of points

Best fit plane without outlier removal. The input points are in 3D coordinates. In practice the points are in a local projected coordinate system, but

Find the blending weight for an image. This is 1 inside the lit area and away from any edges or shadows, then linearly decrease to 0 towards such edge

We use a fixed tile size of 2048x2048 pixels here. Although this may not be the native tile size of the ISIS cube, it seems to be much faster to let

Create a blurred copy of an image

Given an image with float pixels, find the pixels where the image value is non-positive but some of its neighbors have positive values. Create an imag

Partition tile raster files into blocks for parallel mosaicking. Reads the distributed tile list, collects existing files with the given suffix

See the .h file for the documentation.

Build tracks from pairs

Calc albedo given the intensity. albedo = (intensity - haze[0]) / nonlin_ref. See also calcSimIntensity().

Calculate a set of anchor points uniformly distributed over the image Will use opt.num_anchor_points_extra_lines. We append to weight_vec and other qu

Automatic projection determination

Use RANSAC to find the best plane

Based on the current camera models, calculate the world-to-camera transforms. For pinhole cameras, fetch them directly from the camera models. For adj

Find the clamped signed distance to the boundary. Here it is assumed that there is a region in which lit_grass_dist is positive, while in its compleme

Calc the weight for option --curvature-in-shadow-weight

Convert a translation in projected coordinates that aligns the source to the reference to a rotation + translation transform in ECEF. Make use of the

Like PM::ICP::filterGrossOutliersAndCalcErrors, except comparing to a DEM instead. - The point cloud is in GCC coordinates with point_cloud_shift subt

Calculate exposure and haze, and decide if an image should be skipped.

Add poses for the extra desired images based on interpolation, extrapolation, and/or the rig transform.

This function computes / saves measured and simulated intensities, and handles related estimations. If opt.allow_borderline_data is true, create for e

Compute mean and standard deviation of two images. Do it where both are valid.

This is needed to allocate enough storage for the distortion parameters.

This is called by parallel_bundle_adjust just once to accumulate all stats that were done by individual processes.

Calculate the optical offsets for each image. For pinhole and csm frame cameras, read these from the camera model. For other cameras, use half of the

Mapproject interest points onto a DEM and find the norm of their disagreement in meters. It is assumed that dem_georef was created by bilinear interpo

Calculate a few properties of the plane fitted to the given points and print them out

Compute the best-fitting projective transform that maps a set of 3D ground points to 2D image points. See also: applyProjTrans().

Given all camera-to-world transforms, find the median rig transforms. This is robust to outliers. Must handle both frame and linescan cameras for both

Compute and save the covariances