6.4 Complex Methodology for Reverse Engineering …
559
It is necessary to remember that these frames shall be acquired within a single analysis
cycle. This is due to the fact that after a pause in receiving frame flow, a change in
the parameters of illumination and offset, which are not perceived by the organs of
visual perception of the human operator, takes place.
The method described above makes it possible to align a group of frames j with a
fairly high precision; however, the alignment error will not be completely eliminated.
In order to minimize the effect of the error i of alignment, it is usually necessary to
introduce additional alignment methods using a prior information of the analyzed
image. For example, it is possible to use so-called vector guides to obtain panoramic
images of circuit topologies. Unfortunately, no single method ensured high precision
of alignment at the moment of publication of this book, even if all known control
methods were used together.
In this case, the expert recommends to promptly form a specific panorama of
several dozens (or even hundreds) fields of view frame-by-frame (including dozens
of megapixels), reflecting the entire area of the actual microstructure, which in certain
cases can reach values of several square centimeters.
6.4.5 Description of the Process of Layer-by-Layer
Overlapping of Chip Topology Layers
As noted above, when recovering the principal circuit of the analyzed chip, bitmap
images of several main layers of the topology are usually used as source information. Automatic alignment of frames even within a single horizontal layer usually
does not ensure proper alignment between separate layers. Both systematic errors
occurring during photography of a certain layer and usually manifested as angle
rotations and changes in the scale of photography make their contributions. All these
changes become obvious after overlapping several successive layers of an image.
These differences between separate layers affect precise search of coordinates of
elements located in these different layers of the chip image; therefore, it is necessary
in each specific case of the analysis to set the scale and angle of one image according
to the reference image.
All options of image alterations identified in practice can be implemented using
only several basic operations: transfer (movement), rescaling (zooming in or out),
or turning of the image (the terms “rotation” and “re-orientation” are also used)
(Fig. 6.36).
Unfortunately, widely popular correlation methods perform very poorly in practical tasks of layer-by-layer alignment. This is due to the effects of a significant
distortion of the pattern of distribution of the brightness level of the signals on the
examined layer compared to the brightness level of the original image. As of the
moment of publication of this book, there are many known methods of image alignment that were designed considering special limitations in terms of composition and
559
It is necessary to remember that these frames shall be acquired within a single analysis
cycle. This is due to the fact that after a pause in receiving frame flow, a change in
the parameters of illumination and offset, which are not perceived by the organs of
visual perception of the human operator, takes place.
The method described above makes it possible to align a group of frames j with a
fairly high precision; however, the alignment error will not be completely eliminated.
In order to minimize the effect of the error i of alignment, it is usually necessary to
introduce additional alignment methods using a prior information of the analyzed
image. For example, it is possible to use so-called vector guides to obtain panoramic
images of circuit topologies. Unfortunately, no single method ensured high precision
of alignment at the moment of publication of this book, even if all known control
methods were used together.
In this case, the expert recommends to promptly form a specific panorama of
several dozens (or even hundreds) fields of view frame-by-frame (including dozens
of megapixels), reflecting the entire area of the actual microstructure, which in certain
cases can reach values of several square centimeters.
6.4.5 Description of the Process of Layer-by-Layer
Overlapping of Chip Topology Layers
As noted above, when recovering the principal circuit of the analyzed chip, bitmap
images of several main layers of the topology are usually used as source information. Automatic alignment of frames even within a single horizontal layer usually
does not ensure proper alignment between separate layers. Both systematic errors
occurring during photography of a certain layer and usually manifested as angle
rotations and changes in the scale of photography make their contributions. All these
changes become obvious after overlapping several successive layers of an image.
These differences between separate layers affect precise search of coordinates of
elements located in these different layers of the chip image; therefore, it is necessary
in each specific case of the analysis to set the scale and angle of one image according
to the reference image.
All options of image alterations identified in practice can be implemented using
only several basic operations: transfer (movement), rescaling (zooming in or out),
or turning of the image (the terms “rotation” and “re-orientation” are also used)
(Fig. 6.36).
Unfortunately, widely popular correlation methods perform very poorly in practical tasks of layer-by-layer alignment. This is due to the effects of a significant
distortion of the pattern of distribution of the brightness level of the signals on the
examined layer compared to the brightness level of the original image. As of the
moment of publication of this book, there are many known methods of image alignment that were designed considering special limitations in terms of composition and