catalog.txt

来自「NIST Handwriting OCR Testbed」· 文本 代码 · 共 133 行

 append_pi_list - appends the contents of the second list of phrase
                  indices to the members of the first phrase index list.
 append_pi_list_Rel2 - appends the contents of the second list of phrase
                   indices to the members of the first phrase index
                   list using dynamic allocation.
 build_pi_lists - builds lists of phrase indices referencing segmented
                  characters sorted into reading order based on
                  segmentor statistics.
 build_pi_lists_Rel2 - builds lists of phrase indices referencing segmented
                  characters sorted into reading order based on
                  segmentor statistics using dynamic allocation.
 find_bubble_max_y - computes the maximum height of the bubble
                     given the line parameters for b = L(a).
 nearest_nbor_in_right_bubble - incrementally grows a bubble to the
                                right, retuning the nearest blob once
                                blobs are detected within the bubble.
 get_points_within_right_box - initially constrains the search space
                               to the rectangular region bounding the
                               largest possible bubble.
 get_list_in_right_bubble - searches the area within the current bubble
                            for the presence of blobs.
 is_in_right_bubble - determines if the specified point is within the
                      current bubble.
 nearest_nbor_in_left_bubble - incrementally grows a bubble to the
                               left, retuning the nearest blob once
                               blobs are detected within the bubble.
 get_points_within_left_box - initially constrains the search space
                              to the rectangular region bounding the
                              largest possible bubble.
 get_list_in_left_bubble - searches the area within the current bubble
                           for the presence of blobs.
 is_in_left_bubble - determines if the specified point is within the
                     current bubble.
 get_closest_pi_list_to_right - determines which list of phrase indices
                        is closest to the right of another specified
                        list of phrase indices.
 get_closest_pi_list_to_left - determines which list of phrase indices
                        is closest to the left of another specified
                        list of phrase indices.
 current_pi_list - determines if the next blob should be added to the
                   current phrase index list or be added to a new
                   phrase index list.
 current_pi_list_Rel2 - determines if the next blob should be added to the
                   current phrase index list or be added to a new
                   phrase index list using dynamic allocation.
 find_pi_lists - given a list of segmented character centers, sorts the
                 blob indices into phrases according to proper reading
                 order.
 find_pi_lists_Rel2 - given a list of segmented character centers, sorts
                 the blob indices into phrases according to proper reading
                 order using dynamic allocation.
 look_left - initializes the position of a new phrase by using the
             bubbles to look left of the specified position, attempting
             to begin the new phrase at the beginning of a line.
 pi_list_to_right - merge heuristic which determines if one list of
                    phrase indices is to the right of another list.
 pi_list_along_line - merge heuristic which determines if a list of
                      phrase indices lies sufficiently close to a line
                      projection defined by the given parameters.
 merge_pi_lists - collapses lists of phrase indices into complete lines
                  of text by applying two merge heuristics controlled
                  by the provided parameters.
 merge_pi_pair - appends the contents of two lists of phrase indices,
                 removes the merged list from the input lists, and
                 sorts the appended list's members on blob center x's.
 merge_pi_lists_Rel2 - collapses lists of phrase indices into complete
                  lines of text by applying two merge heuristics controlled
                  by the provided parameters using dymanic allocation.
 merge_pi_pair_Rel2 - appends the contents of two lists of phrase indices,
      removes the merged list from the input lists, and sorts the appended
      list's members on blob center x's using dynamic allocation.
 new_pi_list - allocates and initializes a new list of phrase indices,
               storing the top-leftmost blob's index into it.
 new_pi_list_Rel2 - allocates and initializes a new list of phrase indices,
               storing the top-leftmost blob's index into it using dynamic
               allocation.
 build_phrase_map - generate an image map of the central mass of each line of
                    of handprint.
 phrases_from_map - sort blobs into phrase lines using the map.

 norm_small_tall_lists - sort blobs into categories of normal, small, and tall.

 pi_lines_from_map - generate lists of blob indices sorting the blobs
 should_split_tall - test to determine if a tall character should be split
                     across multiple lines of handprint.
 tst_phrase_overlap - tests the amount of character overlap with a line in the
                      phrase map.
 split_across_lines - horizontally splits a character across multiple lines.

 find_closest_line_in_map - finds the closest line in the phrase map to a
                      specified blob.
 hist_blob_colors - generates a histogram from the black pixels in a blob
                    that overlap 0 or more lines in the phrase map.
 draw_phrase_map - actually construct the phrase map image.

 short_phrases_to_problems - removes and appends too short phrases to the
                   problem list.
 phrases2hypstr2 - appends the classes referenced in lists of phrase
                   indices into one long character string.
 phrases2constr2 - appends the confidences referenced in lists of phrase
                   indices into one long character string.
 add_pi_list - adds blob index to specified list dynamically reallocating the
               index list if necessary.
 pi_list_to_blobls - converts the indirection in a list of blob indices
                     into a list of blobs.
 process_nearest_pi - adds the next blob to the current phrase index
                      list and adjusts the bubble location statistics.
 process_nearest_pi_Rel2 - adds the next blob to the current phrase index
                      list and adjusts the bubble location statistics
                      using dynamic allocation.
 remove_from_pi_lists - removes the specified list of phrase indices
                        from the given lists of indices.
 sort_pi_lists_on_y - sorts lists of phrase indices based on the
                      y coordinate of their first blob's center in
                      ascending order.
 sort_pi_list_on_x - sorts the elements of a list of phrase indices
                     based on the x coordinate of each blob's center in
                     ascending order.
 sort_pi_lists_by_length - sorts lists of phrase indices based on the
                           number of elements (blobs) in each list in
                           decending order.
 spell_phrases2 - takes a lists of phrase indices whose elements are
                  sorted into reading order and corrects errors and
                  identifies word boundaries using the specified
                  dictionaries.
 spell_phrases_Rel2 - takes a lists of phrase indices whose elements are
                  sorted into reading order and corrects errors and
                  identifies word boundaries using the specified
                  dictionaries (passed word list rather than dictionary
                  file).
 find_top_left - determines the top-leftmost blob of all the blobs not
                 yet assigned any phrase structure.