In this chapter, we discuss the results and compare the results to the desired output.
5.1 Result Analysis
We were able to obtain the modified RAW image with the synthetic pattern embedded
onto it after executing our Sample Code.Both the RAW images were also converted in
JPEG images using ARGUS sample application.The local statistics that were considered
for comparison were local histogram, contrast, mean luminance, mean intensity and standard
deviation. To give the user a better idea of the dissimilar blocks present in the
modified image, we displayed them in the form of a heatmap where the dissimilar blocks
where represented as cold regions. We also verified the global statistics for the both images
by making use of SSIM .The frames collected before and after the synthetic pattern was
embedded onto them are shown in the following pictures, for indoor lighting conditions:
5.1.1 Different Embedding Frameworks:
- (Simple User Input):We have presented the frames to which the synthetic pattern
was embedded on the location of (100,100). The pattern that is embedded is highlighted.
| (a) Input Image |
(b) Output RAW Image (100,100) |
 |
 |
Figure 5.1: Simple User Input
- (Best suited Block): We have presented the frames to which the synthetic pattern
was embedded on the location of (1212,707) where best suited block is located.
| (a) Input Image |
(b) Output RAW Image (1212,707) |
 |
 |
Figure 5.2: Embedding the synthetic pattern onto the best matched block in terms of
contrast
- (Matching the Mean Contrast):We have presented the frames to which the synthetic
pattern was embedded on the location of (100,100).
| (a) Input Image |
(b) Output RAW Image (100,100) |
 |
 |
Figure 5.3: Matching the contrast of the selected block by adjusting the mean intensity
of the synthetic pattern
- (Matching the Mean Intensity): We have presented the frames to which the
synthetic pattern was embedded on the location of (100,100).
| (a) Input Image |
(b) Output RAW Image (100,100) |
 |
 |
Figure 5.4: Matching the mean intensity of the selected block by adjusting the contrast
of the synthetic pattern
Table 5.1: Comparison of SSIM scores of the Embedding Frameworks
| Embedding Framework |
SSIM Score |
| Simple User Input |
0.9973 |
| Contrast of Pattern |
1.0 |
| Contrast of block |
0.9988 |
| Mean Intensity of block |
0.9990 |
The table above compares the SSIM scores of the various embedding frameworks that
the GUI makes use of. High similarity is shown across a range of characteristics in
the examined picture comparison utilizing SSIM scores and the embedding architecture.
Strong similarity is demonstrated by the remarkably high SSIM score of 0.9973 for simple
user input. It is clear that the option two that is embedding based on the contrast of
the pattern is the most efficient option with the SSIM score of 1.0, so as to be able
to fully preserve the natural image statistics.Furthermore, there is a notable similarity
between the contrast and mean intensity comparisons, with scores of 0.9988 and 0.9990,
respectively, indicating almost comparable characteristics.Overall, the findings show a
strong and constant similarity between the photos on a variety of measures, confirming
their integrity and close likeness within the assessed limits.
5.1.2 Local Statistics Comparison:
A heatmap generated using the ’hot’ colormap and ’nearest’ interpolation is typically
used to visualize dissimilarity or intensity values between pixels or elements in an array.
White often symbolizes the hottest (highest) values in a ’hot’ colormap display. In the
context of picture dissimilarity, this might refer to places where the patterns are highly
similar or have very low dissimilarity to the surrounding regions. In the context of picture
dissimilarity, orange zones may represent places where the embedded patterns or elements
are fairly similar or dissimilar to their surroundings.
| (a) Local Histogram |
(b) Contrast |
 |
 |
| (c) Mean Intensity |
(d) Standard Deviation |
 |
 |
| (e) Mean Luminance |
 |
Figure 5.5: Heat-maps representing the dissimilar blocks in terms of local statistics for
CASE 1 (Simple User Input)
| (a) Local Histogram |
(b) Contrast |
 |
 |
| (c) Mean Intensity |
(d) Standard Deviation |
 |
 |
| (e) Mean Luminance |
 |
Figure 5.6: Heat-maps representing the dissimilar blocks in terms of local statistics for
CASE 2 (Best Suited Block)
| (a) Local Histogram |
(b) Contrast |
 |
 |
| (c) Mean Intensity |
(d) Standard Deviation |
 |
 |
| (e) Mean Luminance |
 |
Figure 5.7: Heat-maps representing the dissimilar blocks in terms of local statistics for
CASE 3 (Match the Contrast)
| (a) Local Histogram |
(b) Contrast |
 |
 |
| (c) Mean Intensity |
(d) Standard Deviation |
 |
 |
| (e) Mean Luminance |
 |
Figure 5.8: Heat-maps representing the dissimilar blocks in terms of local statistics for
CASE 4 (Match the Intensity)