5.17.1.A.ii. Resampling

After the georeferencing process, we may find that the pixels have been oriented differently than the way they were present in the original image coordinate system. Resampling is the process of interpolation the new DN values of the displaced pixels (new pixel location) in the new coordinate system.

Resampling occurs after the image has been georeferenced. This means the image's pixels now correspond to real-world coordinates, but their arrangement may be different from how they were originally organized in the image. As a result, some pixels need to be relocated to fit the new coordinate system. Resampling involves recalculating the values (known as Digital Number or DN values) for these relocated pixels. It effectively 'interpolates' values in a way that maintains the general characteristics of the original image while adapting to the new layout.

Imagine you have a jigsaw puzzle that you have completed, but now you need to fit it into a different-shaped frame. The pieces might need to be resized or adjusted to fit the new frame while still representing the original image as closely as possible.

Three methods of resampling are commonly used, as given below (Figure 5.39). (a) Nearest Neighbour: In this method, the attribute value of the original pixel nearest to a pixel in the output image is assigned to the corresponding cell. (b) Bilinear Interpolation: It assigns the value to a pixel in the output image by taking weighted average of the surrounding four pixels in the original grid nearest to it. (c) Cubic Convolution: It assigns the value to a pixel in the output image by taking weighted average of the surrounding sixteen pixels in the original grid nearest to it.

There are three main techniques employed for resampling:

1. Nearest Neighbour: This method is the simplest. It assigns the value of the nearest pixel from the original image directly to the new pixel location. While it’s less computationally intensive, it may lead to a blocky image since it doesn't smooth the variations between pixel values.
2. Bilinear Interpolation: This technique improves upon the nearest neighbour by considering the values of the four nearest pixels. It calculates a weighted average based on the proximity of these pixels to the new location. This creates smoother transitions between pixel values, reducing blockiness.
3. Cubic Convolution: This is an even more advanced technique where a weighted average of the sixteen nearest pixels is calculated. This results in the highest quality image with smooth transitions but requires more processing power and time.

Think of resampling like adjusting a painting when placed in a new frame. If it's a simple frame (like nearest neighbour), you might only consider the corner pieces of the painting that are in the closest range to fit, creating an angular effect. With a more sophisticated frame (bilinear), you adjust four corner points to ensure a smoother edge. With the most ornate frame (cubic convolution), you adjust sixteen points, producing a seamless transition that looks much better in its new display.

Among the three methods, nearest neighbour is a preferred method as it doesn’t alter the values of the original grid cells assigned to the resampled grid cells but it produces a blocky image. The cubic convolution on the other side does change the values but is more accurate. It generates a smoother image.

Choosing the appropriate resampling method often depends on the specific needs of your analysis or application. The nearest neighbour method, while keeping the original pixel values intact, may not be ideal for giving a clear visual representation due to its tendency to produce a blocky look. On the other hand, cubic convolution, although it modifies pixel values to create a smoother appearance, may be preferred in situations where image quality is critical, such as detailed analysis or presentations.

Consider an artist altering the resolution of an artwork for printing. Using a basic resizing tool (nearest neighbour) keeps the original colors but may create a pixelated effect. However, using an advanced painting software (cubic convolution) to smooth out edges improves the overall look, making the image more visually appealing, even though it slightly alters the colors.

--