A Novel Multi-focus Image Capture and Fusion System for Macro Photography - Advances in Image and Graphics Technologies - page 20

of macro images taken at different focus distances from a photographic subject, with
high precision. After multiple photos are captured, user also needs to fuse these pho-
tos to a sharp one. Multi-scale transforms are often used to analyze the information
content of images for the purpose of fusion, and various methods based on multi-scale
transforms have been proposed in literatures, such as Laplacian pyramid based me-
thods, gradient pyramid based methods and discrete wavelet based methods [2-8].
The basic idea of multi-scale transform is to perform a multi-resolution decomposi-
tion on each source image, then integrate all these decomposition coefficients to
produce a composite representation. The fused image is finally reconstructed by per-
forming an inverse transform. The conventional wavelet-based methods consider the
maximal absolute value of wavelet coefficients or local feature of two images [9-11].
Wavelets are very effective in representing objects with isolated point singularities,
while wavelet bases are not the most significant in representing objects with singulari-
ties along lines. As a consequence, the method based on the wavelet cannot excavate
the edge quality and detail information [12]. The paper proposes a multi-focus image
fusion method based on Laplacian pyramid with a novel weight map selection strategy.
Most multi-focus fusion algorithms for macro photography are assumed that all
source images are point-wise correspondence, that is, the colors at and around any
given pixel in one image correspond to the colors at and around that same pixel in
another image. However, when using the mechanical device to capture different in-
focus images, small motion between adjacent images is inevitable. In these cases,
these images need to be aligned very well before fusion. Otherwise, motion blur effect
will appear in the final fused image. This paper proposes a feature based image
alignment strategy to register the captured images before multi-focus image fusion.
The rest of this paper is organized as follows. The multi-focus image capture de-
vice design is discussed in section 2. The multi-focus image alignment method is
described in section 3. The multi-focus image fusion based on Laplacian pyramids
with weighted maps is given in section 4. After that, experimental results analysis and
evaluation are proposed in section 5. This section also illustrates the proposed macro
photography capture and fusion system with some practical data samples. Finally, the
last section gives some concluding remarks.

2 Macro Photography Capture Device Design

An auto-controlled image capture device is developed to obtain the macro photos for
further processing as shown in figure 1. The device consists of a DSLR Camera with
macro lens; a slide platform to hold the camera moving back and forth; a screw rod to
push the platform; a stepper motor as mechanical power, and a MCU-based control
unit. The camera shutter, flash, and stepper motor are attached to the control unit to
implement the auto-photography system. The thread pitch of the screw rod is 4mm
and with a 360 degree spin, thus the rod pushes the platform to move 4mm. The step
angle of the stepper is 1.8 degree and 8 subdivision controlling scheme is used. There-
fore, with one control pule, the stepper motor spins for 0.225 degree and the platform
moves 0.0025mm. In the experiments, the focal length of the camera is fixed.