Compare segments/laps

Daniel Besse reporter

Preparing for interval comparisons (see #30) wiith stats on interval

→ <<cset 736bba19c887>>

2015-08-04T15:10:03+00:00

Jean-Claude Besse

marked as major

2016-03-29T19:46:21+00:00

Daniel Besse reporter

Some pseudo-algorithm ideas:

Activities search

compute and store for each activity a bounding box
search activities with overlapping bounding box

Segments search

Espace

eqspaced streams in dist for both activities

Resample

for a_idx = [a_act points]:
  a_pos_round[a_idx] = round_to_grid(a_pos[a_idx])
  a_resample = a such that no two consecutive idxs are in the same grid bin

for b_idx = [b_act points]:
  b_pos_round[b_idx] = round_to_grid(b_pos[b_idx])
  b_resample = b such that no two consecutive idxs are in the same grid bin

Categorize

for a_idx = [a_act_resample points]:
  for all 9 neighbors of a_pos_round[a_idx]:
    a_idxs_at[neighbor][] = a_idx

Match

for b_idx = [b_act_resample points]:
  for a_match_idx = a_idxs_at[b_pos_round[b_idx]]:
    cur_seq_len = 0
    while true:
      if isset(seq_len[b_idx][a_match_idx]) :cur_seq_len += seq_len[b_idx][a_match_idx]:
        break
      cur_seq_len += 1
      if end(A) or end(B):
        break
      if !(a_match_idx+cur_seq_len in a_idxs_at[b_pos_round[b_idx+cur_seq_len]])
        break (or allow some short exceptions...)
    seq_len[b_idx][a_match_idx] = cur_seq_len

Filter

for seq = [seq_len]:
  filter according to criteria
  display proposition [start_dist, end_dist]

Segment comparison

[start_dist, end_dist]:

x axis = distance
plot: time difference, speed, power, CdA, etc
x axis = time
map: position

2019-06-07T07:11:36+00:00

Daniel Besse reporter

initial code for segment matching (see #30)

displays common segments in one activity (with loops for instance)
currently only perfect match. Allowing for small discrepancies is TBD (resample may have to be dropped and seq_len directly computed differently for A and B)

→ <<cset 23dac3249979>>