CVJan 13, 2015

An Adaptive Neuro-Fuzzy Inference System Modeling for Grid-Adaptive Interpolation over Depth Images

arXiv:1501.03058v1
Originality Synthesis-oriented
AI Analysis

This work addresses interpolation for depth images, which is incremental as it combines existing fuzzy inference and neural network techniques.

The paper tackled the problem of noise and time-delay in depth image interpolation by applying an Adaptive Neuro-Fuzzy Inference System (ANFIS) to model a four-tap low-pass filter, achieving a general form for uniform interpolation across integer and sub-pixel locations.

A suitable interpolation method is essential to keep the noise level minimum along with the time-delay. In recent years, many different interpolation filters have been developed for instance H.264-6 tap filter, and AVS- 4 tap filter. The present work uses Adaptive Neuro-Fuzzy Inference System (ANFIS) technique to model and investigate the effects of a four-tap low-pass tap filter (Grid-adaptive filter) on a hole-filled depth image. The work demonstrates the general form of uniform interpolations for both integer and sub-pixel locations in terms of the sampling interval and filter length of depth-images via diverse finite impulse response filtering schemes. The demonstrated model combined modelling function of fuzzy inference with the learning ability of artificial neural network.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes