5
recorded videos were then processed with the image stitch-
ing algorithm, and comparisons were made based the image
stitching performance and the visibility of geologic features
and lithology change.
Examples of borehole video frames under three differ-
ent conditions are shown in Figure 7. They were captured at
the similar location within the borehole. However, the side-
view borescope can only cover one side of the borehole, it
is difficult to ensure that the scoping follows the same route
and faces the same direction when recorded these three
videos and minor differences may be observed. In general,
it can be found that the presence of water or dust affects
the appearance of borehole wall. The washed and dried
borehole in Figure 7 (c) can be treated as the ground truth
condition. Compared with Figure 7 (c), there are dusts cov-
ering the borehole wall in Figure 7 (a). Although the dusts
cover the original key points on the borehole wall, there
are areas that were not covered, and the dusts may create
new key points to be detected and matched. Just from the
image stitching point of view, it does not matter whether
key points are on the true borehole wall or introduced from
dusts. At the same time, Figure 7 (b) shows that the pres-
ence of water on the borehole wall causes reflection of the
camera light on part of the borehole image. It potentially
affects the observation of geologic features and introduces
unrealistic key points. The problem with these introduced
key points is that they can change with the movement of
the camera light. Another problem is that the reflections
can be easily blurred with a fast-moving borescope. All
these problems can potentially lead to the presence and the
increased number of stitching failures during the sequential
image stitching with freshly washed boreholes. Although
increasing the difficulties in stitching the borehole images,
the presence of water makes the transition from one layer to
another layer for the laminated shale clearer than the dried
one if comparing Figure 7 (b) and (c).
RESULTS AND DISCUSSIONS
In this section, the image stitching algorithm was used to
generate panoramic borehole images from the borehole
videos recorded in the same borehole under three different
conditions. The influence of borehole wall condition was
compared from the success of image stitching process and
the visibility of geologic features.
The Influence on Image Stitching Performance
One objective of the study was to investigate the influence
of borehole wall conditions on the generation of panoramic
borehole images. However, the sequential image stitching
can also be affected by the borescope movement. The vid-
eos were recorded at different times, and there could be
inconsistent borescope movements during the recordings.
For example, one video could be recorded with constant
and smooth movement, while the other was recorded with
sudden movements at certain points. The ideal condition is
to record the videos with consistent smooth movement and
at a constant speed so that the borehole conditions could
be the only affecting factor for comparison. However, the
borescope movement is hard to be maintained manually,
and as a result, the analysis of the influence of borehole
conditions on the image stitching process was combined
with the analysis of borescope movement.
The question is how to quantify the borescope move-
ment when scoping a borehole. If the distance between the
borescope and borehole wall is assumed to be constant, the
area that one pixel covers on the borehole wall should be
the same on the borehole images. If one borehole video
can be sequentially stitched together to form a panoramic
borehole image, the length of one pixel represented in the
borehole can be approximately calculated by dividing the
measured borehole depth with the total number of pixels
along the borehole depth direction, which is the image
height of the panoramic borehole image. After obtaining
the dimension of area that one pixel covers on the borehole
Figure 7. Borehole video frames under three different
conditions
recorded videos were then processed with the image stitch-
ing algorithm, and comparisons were made based the image
stitching performance and the visibility of geologic features
and lithology change.
Examples of borehole video frames under three differ-
ent conditions are shown in Figure 7. They were captured at
the similar location within the borehole. However, the side-
view borescope can only cover one side of the borehole, it
is difficult to ensure that the scoping follows the same route
and faces the same direction when recorded these three
videos and minor differences may be observed. In general,
it can be found that the presence of water or dust affects
the appearance of borehole wall. The washed and dried
borehole in Figure 7 (c) can be treated as the ground truth
condition. Compared with Figure 7 (c), there are dusts cov-
ering the borehole wall in Figure 7 (a). Although the dusts
cover the original key points on the borehole wall, there
are areas that were not covered, and the dusts may create
new key points to be detected and matched. Just from the
image stitching point of view, it does not matter whether
key points are on the true borehole wall or introduced from
dusts. At the same time, Figure 7 (b) shows that the pres-
ence of water on the borehole wall causes reflection of the
camera light on part of the borehole image. It potentially
affects the observation of geologic features and introduces
unrealistic key points. The problem with these introduced
key points is that they can change with the movement of
the camera light. Another problem is that the reflections
can be easily blurred with a fast-moving borescope. All
these problems can potentially lead to the presence and the
increased number of stitching failures during the sequential
image stitching with freshly washed boreholes. Although
increasing the difficulties in stitching the borehole images,
the presence of water makes the transition from one layer to
another layer for the laminated shale clearer than the dried
one if comparing Figure 7 (b) and (c).
RESULTS AND DISCUSSIONS
In this section, the image stitching algorithm was used to
generate panoramic borehole images from the borehole
videos recorded in the same borehole under three different
conditions. The influence of borehole wall condition was
compared from the success of image stitching process and
the visibility of geologic features.
The Influence on Image Stitching Performance
One objective of the study was to investigate the influence
of borehole wall conditions on the generation of panoramic
borehole images. However, the sequential image stitching
can also be affected by the borescope movement. The vid-
eos were recorded at different times, and there could be
inconsistent borescope movements during the recordings.
For example, one video could be recorded with constant
and smooth movement, while the other was recorded with
sudden movements at certain points. The ideal condition is
to record the videos with consistent smooth movement and
at a constant speed so that the borehole conditions could
be the only affecting factor for comparison. However, the
borescope movement is hard to be maintained manually,
and as a result, the analysis of the influence of borehole
conditions on the image stitching process was combined
with the analysis of borescope movement.
The question is how to quantify the borescope move-
ment when scoping a borehole. If the distance between the
borescope and borehole wall is assumed to be constant, the
area that one pixel covers on the borehole wall should be
the same on the borehole images. If one borehole video
can be sequentially stitched together to form a panoramic
borehole image, the length of one pixel represented in the
borehole can be approximately calculated by dividing the
measured borehole depth with the total number of pixels
along the borehole depth direction, which is the image
height of the panoramic borehole image. After obtaining
the dimension of area that one pixel covers on the borehole
Figure 7. Borehole video frames under three different
conditions