In the following paper, the problem of blocking artefacts caused in Image Compression due to application of Direct Cosine Transforms (DCT) is tackled. DCT is a standard method applied in many image compression algorithms including highly popular JPEG format. A combination of L Filter optimized using Genetic Algorithm is used to solve the problem. This method gave excellent performance on the test images compared to the other commonly used algorithms.
Issues in Image Compression:
Why Image Compression?
Image, by its very nature are extremely bulky in its raw form. Take for example an image from a 7 megapixel camera taking pictures in RGB colour format with 24bits per pixel. Amount of raw memory required for this picture would be 7x106x24 = 168x106 bits = 20MB! This number becomes phenomenal in the case of video, where 30 such images need to be transmitted every second. Hence image compression techniques have become a major research area in today’s information hungry world.
Issues in Image Compression Techniques:
Various compression standards like JPEG (for still images) and MPEG (for video encoding) are currently in use in the multimedia world. Discrete Cosine transform is a common technique used by these standards to compress the image size. One of its major drawbacks is that it suffers from Blocking Artefacts, sometimes so severely, that the entire image is rendered useless. Hence various techniques like L filters are Iterative image restoration applied to the image to improve their quality. The problem with such filters is that they most of them depend on some type of weights or predefined parameters which have to be adjusted for an image on a case by case basis. Modern Soft Computing tools like Neural Networks and Genetic Algorithms have shown excellent performance on selecting the appropriate parameters for the filters in an automated manner.
Print This Post