When you mosaic images, you have overlapping areas. For those overlapping areas, you can specify a cutline so that the pixels on one side of a particular cutline take the value of one overlapping image, while the pixels on the other side of the cutline take the value of another overlapping image. The cutlines can be generated manually or automatically.
When you choose the Set Mode for Intersection, the following features for dealing with image overlap include:
- Loading cutlines from a vector file (a shapefile or arc coverage file)
- Editing cutlines as vectors in the viewer
- Automatic clipping, extending, and merging of cutlines that cross multiple image intersections
- Loading images and calibration information from triangulated block files as well as setting the elevation source
- Selecting mosaic output areas with ASCII files containing corner coordinates of sheets that may be rotated. The ASCII import tool is used to try to parse ASCII files that do not conform to a predetermined format.
- Save cutlines and intersections to a pair of shapefiles
- Loading clip boundary output regions from AOI or vector files. This boundary applies to all output regions. Pixels outside the clip boundary will be set to the background color.
Set Overlap Function
When you are using more than one image, define how they should overlap. Use Set Overlap function to define whether no cutline exists, and if a cutline does not exist, how to handle the overlap of images as well as if a cutline exists, then what smoothing or feathering options to use concerning the cutline.
No Cutline Exists
When no cutline exists between overlapping images, choose how to handle the overlap:
When a cutline does exist between images, decide on smoothing and feathering options to cover the overlap area in the vicinity of the cutline. Use Smoothing Options to choose both the Distance and the Smoothing Filter. The Feathering Options are No Feathering, Feathering, and Feathering by Distance. If you choose Feathering by Distance, type a specific distance.
Automatically Generate Cutlines For Intersection
The current implementation of Automatic Cutline Generation is geometry-based. The method uses the centerlines of the overlapping polygons as cutlines. While this is a very straightforward approach, it is not recommended for images containing buildings, bridges, rivers, and so on because of the possibility the method would make the mosaicked images look obviously inaccurate near the cutline area. For example, if the cutline crosses a bridge, the bridge may look broken at the point where the cutline crosses it.
Weighted Cutline Generation
When your overlapping images contain buildings, bridges, rivers, roads, or anything else where it is very important that the cutline does not break, use Weighted Cutline Generation option. The Weighted Cutline Generation option generates the most nadir cutline first. The most nadir cutline is divided into sections where a section is a collection of continuous cutline segments shared by the two images. The starting and ending points of these sections are called nodes. Between nodes, a cutline is refined based on a cost function. The point with the smallest cost is picked as a new cutline vertex.
Cutline Refining Parameters
In Cutline Refining Parameters section, choose the Segment Length, which specifies the segment length of the refined cutline. The smaller the segment length, the smaller the search area for the next vertex is, and the chances are reduced of the cutline cutting through features such as roads or bridges. This is an especially important consideration for dense urban areas with many buildings. Smaller segment lengths usually slow down the finding of edges, but at least the chances are small of cutting through important features.
Bounding Width specifies the constraint to the new vertices in the vertical direction of the segment between two nodes. More specifically the distance from a new vertex to the segment between the two nodes must be no bigger than the half of the value specified by the Bounding Width field.
Cost Function Weighting Factors
Cost Function used in cutline refinement is a weighted combination of direction, standard deviation, and difference in gray value. The weighting is in favor of high standard deviation, a low difference in gray value, and direction that is closest to the direction between the two nodes. The default value is one for all three weighting factors. At default value, all three components play the same role. When you increment one weighting factor, that factor plays a larger role. If you set a weighting factor to zero, the corresponding component does not play a role at all. If you set all three weighting factors to zero, the cutline refinement is not done, and the weighted cutline is reduced to the most nadir cutline.
Geometry-based Cutline Generation
Geometry-based Cutline Generation is a bit more simplistic because it is based only on the geometry of the overlapping region between images. Pixel values of the involved images are not used. For an overlapping region that only involves two images, the geometry-based cutline can be seen as a center line of the overlapping area that cuts the region into two equal halves. One half is closer to the center of the first image, and the other half is closer to the center of the second image. Geometry-based Cutline Generation runs very quickly compared to Weighted Cutline Generation. Geometry-based generation does not have to factor in pixels from the images. Use the geometry-based method when your images contain homogeneous areas like grasses or lakes, but use Weighted Cutline Generation for images where the cutline cannot break such as buildings, roads, rivers, and urban areas.