6.5 Methods for Restoring Electrical Circuit …
591
6.5.4 Basic Requirements for the Quality of Source Bitmap
Images of the Topology
Reverse-engineering specialists know it well that even in case of compliance with all
the requirements for entering characteristic rectangles, occurrence of various recognition errors is always possible. For example, the application of the described methods
to different source materials demonstrated that the algorithm recognizes 95% of the
circuit elements with a high probability; 5% of defects remain unrecognized due to
defects occurred during preparation (Fig. 6.63).
The necessary condition for performing topology and circuitry transformations is
the quality of the source image of chip topology, which shall be bright and contrast.
If the image contrast is insufficient, automation of the element arrangement process
is nearly impossible to implement. For example, on one of the image fragments
with insufficient contrast, 40 of 40 topology elements remained unrecognized. With
increased contrast, 10 out of 40 analyzed elements remained unrecognized. With
additional adjustment of brightness and contrast, 40 elements were recognized.
Figure 6.64 presents a bitmap image of the topology and brightness distribution
for three vector colors, as well as results of recognition with the specified parameters. Another important factor is the number of pixels per smallest image element.
When analyzing the entered information, the program introduces its errors in the
process of drawing vector elements, which causes their displacement and incorrect
operation of the program (Fig. 6.65).
In order to eliminate the effect of these errors, the number of pixels per smallest
element of the bitmap image shall be at least 10. This number is set during the stage
of acquisition of topology images with the help of digital microscopy means.
When creating the recognition library, it is necessary to input characteristic rectangles so that they reflect the uniqueness of each topology element. If the input of
rectangles is not unique, separate elements are recognized incorrectly (Fig. 6.66).
Since library elements have different dimensions, they are sorted by dimensions
in the program before beginning of the recognition cycle in order to increase the
speed of recognition.
During analysis of elements of various types, situations are possible when a
smaller element is recognized in a larger one (Fig. 6.67), which can lead to incorrect
interpretation of the electrical circuit.
In order to eliminate errors of recognizing small elements and elements with
similar topology, it is necessary to enable an additional element search cycle by
increasing strictness of recognition. An increase in the strictness of recognition for a
number of elements consists in the creation of a subset of elements (group of analogs)
in the recognition library. During analysis of the elements included in this group, an
additional comparison cycle is started. During this cycle, information about position
and brightness of contacts of the library elements are used for element recognition
in addition to information about characteristic rectangles, which ensures increasing
the accuracy of recognition.
591
6.5.4 Basic Requirements for the Quality of Source Bitmap
Images of the Topology
Reverse-engineering specialists know it well that even in case of compliance with all
the requirements for entering characteristic rectangles, occurrence of various recognition errors is always possible. For example, the application of the described methods
to different source materials demonstrated that the algorithm recognizes 95% of the
circuit elements with a high probability; 5% of defects remain unrecognized due to
defects occurred during preparation (Fig. 6.63).
The necessary condition for performing topology and circuitry transformations is
the quality of the source image of chip topology, which shall be bright and contrast.
If the image contrast is insufficient, automation of the element arrangement process
is nearly impossible to implement. For example, on one of the image fragments
with insufficient contrast, 40 of 40 topology elements remained unrecognized. With
increased contrast, 10 out of 40 analyzed elements remained unrecognized. With
additional adjustment of brightness and contrast, 40 elements were recognized.
Figure 6.64 presents a bitmap image of the topology and brightness distribution
for three vector colors, as well as results of recognition with the specified parameters. Another important factor is the number of pixels per smallest image element.
When analyzing the entered information, the program introduces its errors in the
process of drawing vector elements, which causes their displacement and incorrect
operation of the program (Fig. 6.65).
In order to eliminate the effect of these errors, the number of pixels per smallest
element of the bitmap image shall be at least 10. This number is set during the stage
of acquisition of topology images with the help of digital microscopy means.
When creating the recognition library, it is necessary to input characteristic rectangles so that they reflect the uniqueness of each topology element. If the input of
rectangles is not unique, separate elements are recognized incorrectly (Fig. 6.66).
Since library elements have different dimensions, they are sorted by dimensions
in the program before beginning of the recognition cycle in order to increase the
speed of recognition.
During analysis of elements of various types, situations are possible when a
smaller element is recognized in a larger one (Fig. 6.67), which can lead to incorrect
interpretation of the electrical circuit.
In order to eliminate errors of recognizing small elements and elements with
similar topology, it is necessary to enable an additional element search cycle by
increasing strictness of recognition. An increase in the strictness of recognition for a
number of elements consists in the creation of a subset of elements (group of analogs)
in the recognition library. During analysis of the elements included in this group, an
additional comparison cycle is started. During this cycle, information about position
and brightness of contacts of the library elements are used for element recognition
in addition to information about characteristic rectangles, which ensures increasing
the accuracy of recognition.