Automated Removal of Image Compression Artifacts Improves an algorithm proposed by Luo and Ward for the removal of edges of the blocks in regions that are affected by blockiness effects.
DCT compression methods can introduce visible artifacts at block boundaries of images. In this project we take a look at an algorithm proposed by Luo and Ward for the removal of edges of the blocks in regions that are affected by blockiness effects. The algorithm uses an adaptive approach where it considers the local information of the image before proceeding with the modification of pixel values based on weighted average of neighbouring blocks. Experiments show that the algorithm can significantly reduce the visible blockiness artifacts from the given images.
The aim of our project is to create an algorithm to extract a pedestrian path from the rest of the image. The target is achieved by using image segmentation by K-means and developing a code to implement it in matlab. The idea was to differentiate the path from the rest of the image in terms of colour as the rest of the objects are turned white so that the path is differentiated. The experiments have shown promising results although there is some room for improvement. The quality of segmentation can be improved by various techniques and will automatically improve the detail in our output images as well.
See Paper for more details: https://github.com/asjad99/Image-Processing/blob/master/Paper.pdf