BMP-filters

This repository contains the program that allows you to manipulate files ending in ".bmp" by using a selection of filters that can be applied to the original image to create an output.

Filters

Grayscale: Takes an input image and turns the image into a black-and-white by averaging the red, green and blue (RGB) values for a given pixel, which has the additional benefit of retaining the original image brightness.

Reflection: Takes an input image and reflects the image horizontally. You would get the resulting image by placing the original image in front of a mirror.

Box Blur: Takes an input image and implements a box blur. Box blur is implemented by taking each pixel and, for each colour value, giving it a new value by averaging the colour values of the neighbouring pixels where applicable.

Sobel Operator: Takes an input image and applies the Sobel operator on the given image. Like Box Blur, edge detection works by modifying each pixel based on a 3x3 matrix surrounding a given pixel. However, in this scenario, we use the Sobel operator to compute the new value of each pixel by taking a weighted sum of the values for the surrounding pixels. Since edges can occur in both x and y directions within an image, we use two different "kernels" to compute the edges in a particular direction for each colour channel for a given pixel. Each channel for a given pixel will be processed using the Gx and Gy kernels, meaning each pixel will have six(6) values.

If 'I' is the source image, then we can apply the kernels Gx and Gy to each pixel within an image:

$$Gx=\begin{pmatrix} +1&0&-1\\ +2&0&-2\\ +1&0&-1&\end{pmatrix} * I $$

$$Gy=\begin{pmatrix} +1&+2&+1\\ 0&0&0\\ -1&-2&-1&\end{pmatrix} * I $$

To combine the values for a given channel after using the kernels Gx and Gy, we can:

$$O=\sqrt{Gx^2 + Gy^2}$$

which generates 'O' the output for a given pixel channel, which squares the result for each kernel, then adds that result and finally takes the square root to normalise the result.

If the result exceeds 255, the C function fmin is used to cap the output at 255.

Compilation

A GNU Make, Makefile has been provided. To compile the program, clone this repository, enter the directory, then:

mahdi@malkawimahdi:~$  make filter

Usage

The program generates an output depending on the flag used in the command line.

To apply grayscale:

mahdi@malkawimahdi:~$  ./filter -g INPUT_FILE.bmp OUTPUT_FILE_NAME.bmp

To apply reflect:

mahdi@malkawimahdi:~$  ./filter -r INPUT_FILE.bmp OUTPUT_FILE_NAME.bmp

To apply box blur:

mahdi@malkawimahdi:~$  ./filter -b INPUT_FILE.bmp OUTPUT_FILE_NAME.bmp

To apply Sobel operator:

mahdi@malkawimahdi:~$  ./filter -e INPUT_FILE.bmp OUTPUT_FILE_NAME.bmp