Calculating the per pixel surface intensity for small kernels has been a continual issue. The basic problem is that when the pixel size becomes comparable to the kernel size, the location of the kernel centroid within a pixel can have a large effect on the surface intensity calculated within that pixel. This leads to an unnormalized surface intensity and a mis-estimation of the richness and other source parameters.
One possible solution would be to subsample the surface intensity within each pixel (or just the central pixels). Whether this is technically and computationally feasible needs to be investigated.
kadrlica
commented
6 years ago
Calculating the per pixel surface intensity for small kernels has been a continual issue. The basic problem is that when the pixel size becomes comparable to the kernel size, the location of the kernel centroid within a pixel can have a large effect on the surface intensity calculated within that pixel. This leads to an unnormalized surface intensity and a mis-estimation of the richness and other source parameters.
One possible solution would be to subsample the surface intensity within each pixel (or just the central pixels). Whether this is technically and computationally feasible needs to be investigated.