Extended Visual Cryptography for Photograph Images - Visual Cryptography and Secret Image Sharing - page 107

4.4 Variations of Photograph Visual Cryptography
Although extended capability is a crucial aspect for visual cryptography with
a photograph image, contrast of resulting images becomes very low ( 4 at maxi-
mum) due to the tradeoff between relative differences of the shadow image and
reconstructed image. Contrast tends to much lower if the number of shadow
images increases. In order to enhance contrast of encrypted images, several
researches assumed other kinds of operations for superimposition, namely,
"Cover," [32] "XOR," [3, 23] "NOT," etc. [40] Sometimes even a certain com-
putation is required for decryption [5, 20, 24]. Most of the approaches are no
longer realized by stacking transparencies. 2 Those schemes may not be cate-
gorized as visual cryptography, because the most important characteristics of
visual cryptography is the capability of visual decryption without any compu-
tation. Therefore, we will not discuss these type of approaches assuming other
operations than Boolean "OR." Instead our discussion will mainly focus on
(2; 2) schemes. There have been a lot of studies that aim at incorporating color
into visual cryptography [27, 35, 39, 22, 44, 12, 13, 7, 30, 36, 46, 19]. These
approaches are strongly related to the techniques for handling continuous-tone
images. However, this chapter will not discuss those color studies because this
book contains a special chapter dedicated to color visual cryptography.
4.4.1 Approaches to Photograph Visual Cryptography
The main issue for incorporating photographs into visual cryptography is the
quality of resulting images, i.e., pixel expansion, relative differences, tone lev-
els, as we discussed in Section 4.3.2. A lot of approaches to photograph visual
cryptography intend to improve the image quality by introducing a certain
limitation and/or by exploiting image processing techniques. Table 4.2 sum-
marizes those approaches. For instance, [4, 15, 37, 14, 6] limit their shadow
image to a random-dot binary image ("rand."). Actually they do not intend ex-
tended visual cryptography [33, 34, 21, 41, 9, 47]. Generate very similar images
("sim.") or a positive/negative pair of images ("p/n") as encrypted shadow
images. Sometimes only logo-like images with the trace of shadows ("logo tr.")
can be reconstructed as a secret image [33, 34, 21, 41, 9, 10, 45]. [10, 25, 42] uti-
lize halftoning techniques to make pixel expansion m = 1. [28, 42, 45, 25, 26]
adjust tone, i.e., dynamic range, of images for improving image quality. Some
studies introduce continuous-tone subpixels into encrypted shadow images to
obtain continuous-tone results [45, 29]. The rest of this section explains those
approaches.
2 Only "Cover" can be physically realized with transparencies and opaque sheets. Physical
implementation of "XOR" is possible by exploiting polarization.