Blur is implemented using four different techniques: box, lens, Gaussian, and motion. The box blur technique is the most straightforward one: each pixel in the resulting image has a value equal tothe average value of its neighboring pixels in the input image. The lens blur is also known as the Bokeh effect, which mimics the the aesthetic quality of the blur produced in out-of-focus parts of an image. It canbe emulated by convolving the image with a kernel that corresponds to the image of an out-of-focuspoint source taken with a real camera. Our lens blur implementation allows user to specify a radius ofpoint source/convolution and form such a kernel to mimic the Bokeh effect. The third method, Gaussian blur, alsoknown as Gaussian smoothing, is the result of blurring an image by using a Gaussian kernel to convolvewith the image pixels. User can choose the size of the Gaussian kernel. The distribution of the Gaussian kernel corresponds to the values of a 3-dimensional Gaussian distribution. The fourth method of blurring is the motion blur,which blurs the image in a specified direction to give the impression of movement. Again, this is achievedby applying a convolution kernel on the image with the goal of emphasizing values in one direction. Usercan specify the radius of kernel and four different angle of motion blur: 0, 45, 90, and 135 degree.
See the project report for more details.