pointmap.py

from helpers import poseRt, hamming_distance, add_ones
from constants import CULLING_ERR_THRES
from frame import Frame
import time
import numpy as np

import json


#from optimize_crappy import optimize

LOCAL_WINDOW = 20
#LOCAL_WINDOW = None

class Point(object):
  # A Point is a 3-D point in the world
  # Each Point is observed in multiple Frames

  def __init__(self, mapp, loc, color, tid=None):
    self.pt = np.array(loc)
    self.frames = []
    self.idxs = []
    self.color = np.copy(color)
    self.id = tid if tid is not None else mapp.add_point(self)

  def homogeneous(self):
    return add_ones(self.pt)

  def orb(self):
    return [f.des[idx] for f,idx in zip(self.frames, self.idxs)]

  def orb_distance(self, des):
    return min([hamming_distance(o, des) for o in self.orb()])

  def delete(self):
    for f,idx in zip(self.frames, self.idxs):
      f.pts[idx] = None
    del self

  def add_observation(self, frame, idx):
    assert frame.pts[idx] is None
    assert frame not in self.frames
    frame.pts[idx] = self
    self.frames.append(frame)
    self.idxs.append(idx)

class Map(object):
  def __init__(self):
    self.frames = []
    self.points = []
    self.max_frame = 0
    self.max_point = 0

  def serialize(self):
    ret = {}
    ret['points'] = [{'id': p.id, 'pt': p.pt.tolist(), 'color': p.color.tolist()} for p in self.points]
    ret['frames'] = []
    for f in self.frames:
      ret['frames'].append({
        'id': f.id, 'K': f.K.tolist(), 'pose': f.pose.tolist(), 'h': f.h, 'w': f.w,
        'kpus': f.kpus.tolist(), 'des': f.des.tolist(),
        'pts': [p.id if p is not None else -1 for p in f.pts]})
    ret['max_frame'] = self.max_frame
    ret['max_point'] = self.max_point
    return json.dumps(ret)

  def deserialize(self, s):
    ret = json.loads(s)
    self.max_frame = ret['max_frame']
    self.max_point = ret['max_point']
    self.points = []
    self.frames = []

    pids = {}
    for p in ret['points']:
      pp = Point(self, p['pt'], p['color'], p['id'])
      self.points.append(pp)
      pids[p['id']] = pp

    for f in ret['frames']:
      ff = Frame(self, None, f['K'], f['pose'], f['id'])
      ff.w, ff.h = f['w'], f['h']
      ff.kpus = np.array(f['kpus'])
      ff.des = np.array(f['des'])
      ff.pts = [None] * len(ff.kpus)
      for i,p in enumerate(f['pts']):
        if p != -1:
          ff.pts[i] = pids[p]
      self.frames.append(ff)

  def add_point(self, point):
    ret = self.max_point
    self.max_point += 1
    self.points.append(point)
    return ret

  def add_frame(self, frame):
    ret = self.max_frame
    self.max_frame += 1
    self.frames.append(frame)
    return ret

  # *** optimizer ***

  # def optimize(self, local_window=LOCAL_WINDOW, fix_points=False, verbose=False, rounds=50):
  #   err = optimize(self.frames, self.points, local_window, fix_points, verbose, rounds)
  #
  #   # prune points
  #   culled_pt_count = 0
  #   for p in self.points:
  #     # <= 4 match point that's old
  #     old_point = len(p.frames) <= 4 and p.frames[-1].id+7 < self.max_frame
  #
  #     # compute reprojection error
  #     errs = []
  #     for f,idx in zip(p.frames, p.idxs):
  #       uv = f.kps[idx]
  #       proj = np.dot(f.pose[:3], p.homogeneous())
  #       proj = proj[0:2] / proj[2]
  #       errs.append(np.linalg.norm(proj-uv))
  #
  #     # cull
  #     if old_point or np.mean(errs) > CULLING_ERR_THRES:
  #       culled_pt_count += 1
  #       self.points.remove(p)
  #       p.delete()
  #   print("Culled:   %d points" % (culled_pt_count))

    return err