sort_region — 根据相对位置对区域进行排序。
sort_region(Regions : SortedRegions : SortMode, Order, RowOrCol : )
算子 sort_region sorts the regions
with respect to their relative position.
All sorting methods with the exception of 'character'
use one point of the region. With the help of the parameter
RowOrCol = 'row' these points will be
sorted according to their row and then according to their
column. By using 'column', the column value will
be used first.
The following values are available for the parameter
SortMode:
The regions are treated like characters, which can be read
horizontally or vertically. They are sorted according
to the reading direction given by RowOrCol with
the following specifications:
'row'：Regarded as rows, the reading direction is firstly from left to right and then from top to bottom.
'column'：Regarded as columns, the reading direction is firstly from top to bottom and then from left to right.
For SortMode 'character', a second numeric value
can be passed to SortMode in addition to 'character'.
This value specifies the maximum percentage of overlap. This overlap
depends on the parameter RowOrCol and is determined as
follows:
RowOrCol = 'row'：overlap of row coordinates
between two regions
RowOrCol = 'column'：overlap of column
coordinates between two regions,
whereas the percentage of overlap is calculated using the smaller one
of the overlapping regions (with respect to the specified axis).
Regions that do not overlap in their coordinates or have an overlap
of less than the set percentage are seen as being in different rows
(or columns, respectively) and sorted accordingly. In case the overlap
is larger, the regions are seen as being in the same row (or column)
instead and therefore sorted within the row (or column). For example
when this additional parameter is set to 0, all regions that overlap
in their row (or column) coordinates are sorted within the same row
(or column). The default value of this parameter is 15 which means
that adjacent rows (or columns) can have an overlap of maximum 15%
in order to be considered as different rows (or columns).
Note that for Order = 'false', the characters
are sorted in a reversed order to the one described above.
The point with the lowest column value in the first row of the region.
The point with the highest column value in the last row of the region.
Upper left corner of the surrounding rectangle.
Upper right corner of the surrounding rectangle.
Lower left corner of the surrounding rectangle.
Lower right corner of the surrounding rectangle.
The parameter Order determines whether the
sorting order is increasing or decreasing: using
'true' the order will be increasing, using
'false' the order will be decreasing.
Regions (输入对象) region-array → object
Regions to be sorted.
SortedRegions (输出对象) region-array → object
Sorted regions.
SortMode (输入控制) tuple(-array) → (string / integer / real)
Kind of sorting.
默认值： 'first_point'
值列表： 'character', 'first_point', 'last_point', 'lower_left', 'lower_right', 'upper_left', 'upper_right'
Order (输入控制) string → (string)
Increasing or decreasing sorting order.
默认值： 'true'
值列表： 'false', 'true'
RowOrCol (输入控制) string → (string)
Sorting first with respect to row, then to column.
默认值： 'row'
值列表： 'column', 'row'
如果参数正确，算子
sort_region 返回值 2 ( H_MSG_TRUE )。否则将抛出异常。
do_ocr_multi_class_mlp,
do_ocr_single_class_mlp
基础