5.17.2.B. Contrast Enhancement

The contrast is defined as the maximum difference in color or intensity between two objects in an image. If the contrast is poor (low), it is impossible to distinguish between various objects and they are perceived as the same object.

Contrast in images is crucial for visually distinguishing different objects. When we say contrast is the maximum difference in color or intensity, we refer to how bright or dark, or how different in color two things appear. If this difference is very small (low contrast), it becomes challenging for the human eye to tell one object from another, making everything blend together.

Imagine you are at a party with dim lighting. If everyone is dressed in similar dark colors, it becomes hard to tell one friend from another. This situation is akin to low contrast in an image. However, if someone wears a bright, vivid outfit, they'll stand out, similar to high contrast in a well-lit image where different objects can be easily distinguished.

Contrast enhancement involves changing the original DN values so that more available range is utilised, thereby increasing the contrast between the objects and their background. It basically improves the interpretability for human viewing, and provides enhanced input to be used for image processing.

The main goal of contrast enhancement is to improve the visibility of features in an image by adjusting the range of colors or intensities displayed. By changing the Digital Number (DN) values, we can stretch the range so features stand out more against their backgrounds. This is essential for both viewers and automated processes that analyze images.

Think of contrast enhancement like adjusting the brightness and contrast on your TV. When you increase the contrast, the images you see become sharper and more defined. This allows you to clearly see details that might have been lost in a flatter image, just like enhancing the details makes objects more recognizable in satellite imagery.

There are many different techniques and methods of enhancing the contrast, and details can be found in Garg (2022). The linear contrast enhancement is the most popular technique used for image enhancement. It involves identifying the minimum and maximum DN values in the image, and applying a linear transformation to stretch the present range to occupy the full range (e.g., 0-255 in an 8-bit image).

Linear contrast enhancement is a straightforward approach where we look at the lowest and highest pixel values in the image. By mapping these two extremes to the full range of pixel values (like from 0 to 255 in an 8-bit image), we effectively brighten light areas and darken dark areas, which improves visibility and differentiation of objects.

Imagine you’re adjusting a watercolor painting's brightness and contrast. If the colors are too faded, you wouldn't be able to appreciate the rich tones. By applying a linear adjustment that spreads out the colors, you bring out the vibrancy of the painting, making individual colors pop and the overall image clearer, similar to enhancing the contrast in images to make them more visually appealing.

The linear contrast enhancement is mathematically represented as-

DN_out = ((DN_in - DN_min) / (DN_max - DN_min)) * No. of grey levels.

To quantitatively describe how linear contrast enhancement works, we can use this formula where DN_out refers to the new pixel value, DN_in is the original pixel value, DN_min and DN_max represent the minimum and maximum pixel values in the original image, and 'No. of grey levels' indicates the total range of values that can be assigned (like 256 for an 8-bit image). This equation adjusts each pixel based on its original value relative to the overall range of pixel values.

Consider cooking pasta. If you only have a small pot of water compared to the amount of pasta you need, the pasta won't cook properly because there's not enough water. When you scale up the amount of water (similar to stretching pixel values), every piece gets the right amount of cooking (contrast enhancement), ensuring each piece is adequately differentiated and identifiable in the final dish.

The main purpose of image enhancement is to increase the contrast in a low contrast image. It does not add any information to the original image but it enhances the visual appearances of already captured features.

Enhancing contrast is not about providing new information; it is about making what is already there more visible. By amplifying the differences in light and dark areas as well as colors, we improve the chances of accurately interpreting the image. However, the essential data remains unchanged, and the enhancement merely makes existing features clearer.

Think of viewing a photograph. Sometimes, photographs can be a bit dim due to poor lighting. If you increase the brightness and contrast on your photo editing app, you reveal details hidden in the shadows or highlights. This process does not create new content; it makes what's already in the photo clearer, which is the essence of image contrast enhancement in remote sensing.

--