Removing grain and lowlevel noise

Because no two pixels are alike, and because photons and electrons are discrete particles, digital images of faint objects are grainy. To a considerable extent, grain can be removed by software without obscuring the desired detail in the image. After all, if you can tell the difference between the grain and the image, then, in principle, so can the computer.

All computer graphics textbooks say that you can reduce noise by blurring the image - that is, by averaging each pixel with some of its neighbors. That is true but unsatisfying for two reasons. First, blurring the image makes stars disappear. Second, if you blur a random grain pattern without reducing the contrast, often what you end up with is just a bigger random pattern.

Figure 14.9. Neat Image removes virtually all the grain from the stacked image in Figure 14.1 (p. 179) without blurring stars.

A better approach is to try to distinguish intelligently between grain and signal. One piece of software that does an impressive job of this is Neat Image (, available both as a standalone package and as a Photoshop plug-in. As Figure 14.9 shows, Neat Image can almost work magic. The noise reduction filter in Photoshop CS2 is much less potent.

In operation, Neat Image attempts to find a 128 x 128-pixel area of the image that is free of stars or other detail. From this, it constructs a mathematical model of the grain. If the reference area does contain stars, you're likely to end up with excessively aggressive despeckling; Neat Image will turn into a star eater.

Because of the Bayer matrix, much of the grain in a DSLR image is actually "color noise" ("chrominance noise"), variation in color rather than in brightness. Neat Image recognizes this and lets you adjust the levels of luminance and chrominance noise correction separately.

The algorithms used in Neat Image are not made public, but the key idea has to be that each pixel is averaged, not with all of its neighbors, but only with those that are not too different from it in the first place. In this way, low-level noise is removed but star images remain sharp. It also appears that Neat Image tries to find and follow straight edges; this is a wise policy in daytime photography, but if an astronomical image is overprocessed in Neat Image, nebulae and even stars begin to turn into linear streaks.

Photoshop Secrets

Photoshop Secrets

Are You Frustrated Because Your Graphics Are Not Looking Professional? Have You Been Slaving Over Your Projects, But Find Yourself Not Getting What You Want From Your Generic Graphic Software? Well, youre about to learn some of the secrets and tips to enhance your images, photos and other projects that you are trying to create and make look professional.

Get My Free Ebook

Post a comment