TITLE:
3D Depth Measurement for Holoscopic 3D Imaging System
AUTHORS:
Eman Alazawi, Mohammad Rafiq Swash, Maysam Abbod
KEYWORDS:
Holoscopic 3D Image, Edge Detection, Auto-Thresholding, Depthmap, Integral Image, Local Histogram Analysis, Object Recognition and Depth Measurement
JOURNAL NAME:
Journal of Computer and Communications,
Vol.4 No.6,
May
30,
2016
ABSTRACT: Holoscopic 3D imaging is a true 3D imaging
system mimics fly’s eye technique to acquire a true 3D optical model of a real
scene. To reconstruct the 3D image computationally, an efficient implementation
of an Auto-Feature-Edge (AFE) descriptor algorithm is required that provides an
individual feature detector for integration of 3D information to locate objects
in the scene. The AFE descriptor plays a key role in simplifying the detection
of both edge-based and region-based objects. The detector is based on a
Multi-Quantize Adaptive Local Histogram Analysis (MQALHA) algorithm. This is
distinctive for each Feature-Edge (FE) block i.e. the large contrast changes
(gradients) in FE are easier to localise. The novelty of this work lies in
generating a free-noise 3D-Map (3DM) according to a correlation analysis of
region contours. This automatically combines the exploitation of the available
depth estimation technique with edge-based feature shape recognition technique.
The application area consists of two varied domains, which prove the efficiency
and robustness of the approach: a) extracting a set of setting feature-edges,
for both tracking and mapping process for 3D depthmap estimation, and b)
separation and recognition of focus objects in the scene. Experimental results
show that the proposed 3DM technique is performed efficiently compared to the
state-of-the-art algorithms.