Region Adaptive, Unsharp Masking Based Lanczos-3 Interpolation for 2-D Up-Sampling: Crisp-Rule Versus Fuzzy-Rule Based Approach

Abstract

Up-sampling problem is very crucial in image communication as it plays an important role in restoring the low resolution 2-D signals at the receiver. Generally, at the transmitting end, a video intra frame is sub-sampled to lessen the bandwidth required for transmission. At the receiver, the resolution of the sub sampled intra frame is improved to the original by a suitable interpolation technique. This process lessens the signal bandwidth for transmission through a communication link and hence avoids channel congestion. Most of the up-sampling scheme based on interpolation generates undesirable blurring artefacts in the up-sampled video intra frame. This results in signal deterioration in terms of loss of fine details and critical edge information. Such problems occur due to the resemblance of the up-sampling process with the low pass filtering operation. In order to resolve this problem, both crisp-rule and fuzzy-rule based hybrid interpolation techniques are proposed here. A crisp-rule based technique (Proposed-1) although gives better results in certain cases but fails to provide considerable performance under varying constraints such as variation in zoom in or zoom out conditions, change in compression ratio and video characteristics. This is basically due to improper mapping by crisp-rule based technique between input and output values. Such problems can be avoided by fuzzy-rule based technique (Proposed-2) which employs a proper and precise mapping between input and output values using fuzzy inference system. This book chapter critically compares the capabilities and limitations of crisp-rule and fuzzy-rule based up-sampling techniques and their relevance in perspective of video scalability, compatibility, complexities, quality enhancement and real time applications.