Levelset method for 4D change detection

.github/workflows/ci.yml

@@ -67,7 +67,7 @@ jobs:
 
     - name: Run Python tests
       run: |
-        python -m pytest --nbval
+        python -m pytest --nbval --ignore=jupyter/levelset-analysis.ipynb
 
   coverage-test:
     name: Coverage Testing
@@ -195,6 +195,8 @@ jobs:
         echo "leak:PyArrayMethod_FromSpec_int" >> supp.txt
         echo "leak:PyDict_Copy" >> supp.txt
         echo "leak:PyUnicode_New" >> supp.txt
+        echo "leak:PyUFunc_FromFuncAndDataAndSignatureAndIdentity" >> supp.txt
+        echo "leak:PyArrayIdentityHash_New" >> supp.txt
         echo "leak:pyo3::types::function::PyCFunction::internal_new_from_pointers" >> supp.txt
         echo "leak:pyo3::types::function::PyCFunction::internal_new::" >> supp.txt
         # hack to prevent external libs from dlclosing libraries,

.gitignore

@@ -177,3 +177,6 @@ dmypy.json
 
 # Cython debug symbols
 cython_debug/
+jupyter/new_data_24_168_1/
+
+jupyter/synthetic_k0_10_10/

jupyter/4dobc-analysis.ipynb

@@ -143,6 +143,24 @@
     "We can plot single objects interactively:"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test = objects[1]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -216,7 +234,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.6"
+   "version": "3.11.9"
   }
  },
  "nbformat": 4,

jupyter/levelset-analysis.ipynb

@@ -0,0 +1,231 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import py4dgeo"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "analysis = py4dgeo.SpatiotemporalAnalysis(\"synthetic.zip\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "analysis.invalidate_results(seeds=False, objects=True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "\n",
+    "Options for the data setup:\n",
+    "- `first_timestep` (int): The first timestep to process. Default is 0.\n",
+    "- `last_timestep` (int): The last timestep to process. If set to -1, all timesteps until the end are processed. Default is -1.\n",
+    "- `timestep_interval` (int): The interval between timesteps to process. Default is 1.\n",
+    "\n",
+    "Example for the interval pairings:\n",
+    "interval: 24 (every timestep will be compared to the timestep + 24)\n",
+    "first_timestep: 0 (start with the 0 index)\n",
+    "last_timestep: 75 (last index to be used will be 75, so the highest index accessed will be 99)\n",
+    "step0: f1:0 f2:24\n",
+    "step1: f1:1 f2:25\n",
+    "step2: f1:2 f2:26\n",
+    "step3: f3:3 f2:27\n",
+    "...\n",
+    "stepN f1:75 f2:99\n",
+    "\n",
+    "Options for level set algorithm:\n",
+    "- `reuse_intermediate` (bool): Re-use intermediate calculations\n",
+    "(neighbors, normals, tangents). Default is True.\n",
+    "- `active_contour_model` (str): Active contours model, either\n",
+    "'chan_vese' or 'lmv' (Local Mean and Variance). Default is 'chan_vese'.\n",
+    "- `num_cycles` (int): Number of cycles, each runs a number of steps\n",
+    "then stores a result. Default is 12.\n",
+    "- `num_steps` (int): Number of steps per cycle. Default is 50.\n",
+    "- `num_smooth` (int): Number of smoothing passes for zeta.\n",
+    "Default is 1.\n",
+    "- `stepsize` (int): Explicit Euler step size. Default is 1000.\n",
+    "- `nu` (float): Controls regularization. Default is 0.0001.\n",
+    "- `mu` (float): Controls curvature. Default is 0.0025.\n",
+    "- `lambda1` (float): Controls zeta-in term. Default is 1.0.\n",
+    "- `lambda2` (float): Controls zeta-out term. Default is 1.0.\n",
+    "- `epsilon` (float): Heaviside/delta approximation \"width\",\n",
+    "is scaled with `h`. Default is 1.0.\n",
+    "- `h` (float): Approximate neighborhood radius\n",
+    "(all k neighbors should be within). Default is 2.5.\n",
+    "- `k` (int): Number of kNN neighbors. Default is 7.\n",
+    "- `tolerance` (float): Termination tolerance. Default is 5e-5.\n",
+    "- `cue_clip_pc` (float): Robust cues, clip at X%. Default is 99.9.\n",
+    "- `vox_size` (int): Initialization voxel size. Default is 10.\n",
+    "- `init_pc` (int): Initialization cue percentage. Default is 50.\n",
+    "- `init_method` (str): Initialization method, either 'voxel' or 'cue'.\n",
+    "Default is 'voxel'.\n",
+    "- `extraction_threshold` (int): Neighbor threshold for points\n",
+    "to be extracted\n",
+    "(must have >= salient neighbors to be extracted).\n",
+    "Calculated as `k // 2`.\n",
+    "- `center_data` (bool): Recenter cues by subtracting cue median.\n",
+    "Default is False.\n",
+    "\n",
+    "\n",
+    "Options for the shape analysis:\n",
+    "- `_filter` (str): Choose between the positive and negative data files.\n",
+    "Default is 'positive'.\n",
+    "- `distance_threshold` (float): How far points can be to still be\n",
+    "considered of the same object. Default is 1.\n",
+    "- `change_threshold` (float): How high the change value needs to be to\n",
+    "be considered a valid entry. Default is 0.5.\n",
+    "- `alpha` (float): Alpha parameter for the alpha shape identification,\n",
+    "the lower the smoother the shape, but less exact. Default is 1.\n",
+    "- `area_threshold` (int): Area threshold for filtering small polygons.\n",
+    "Default is 100.\n",
+    "- `iou_threshold` (float): Intersection over Union (IoU) threshold for\n",
+    "assigning objects IDs in different time steps. Default is 0.5.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "algo = py4dgeo.LevelSetAlgorithm(\n",
+    "    first_timestep=0,\n",
+    "    last_timestep=4,\n",
+    "    timestep_interval=10,\n",
+    "    alpha=0.1,\n",
+    "    iou_threshold=0.5,\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "algo.run(analysis)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "objects = analysis.objects\n",
+    "print(objects)\n",
+    "test_obj = objects[0]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test_obj.plot()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test_obj.change_histogram(nbins_x=10)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Overview of the most important object properties\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test_obj.gdf"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Filter for positiv or negative files used in the ojbect generation\n",
+    "print(\"Filter:\", test_obj.filter)\n",
+    "print(\"Timestep interval:\", test_obj.interval)\n",
+    "print(\"IoU threshold:\", test_obj.iou_thresholds)\n",
+    "\n",
+    "\n",
+    "# The .timesteps attribute is a list of all the timesteps available for this object.\n",
+    "# These indices correspond to the first epoch of the calculation\n",
+    "# These indices are used for the .indicies, .coordinates and .distances attributes\n",
+    "\n",
+    "print(\"Timesteps:\", test_obj.timesteps)\n",
+    "print(\"Indices keys:\", test_obj.indices.keys())\n",
+    "print(\"Coordinates keys:\", test_obj.coordinates.keys())\n",
+    "print(\"distances keys:\", test_obj.distances.keys())\n",
+    "\n",
+    "# print(test_obj.indices[0])\n",
+    "# print(test_obj.coordinates[0])\n",
+    "# print(test_obj.distances[0])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.9"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

pyproject.toml

@@ -39,6 +39,16 @@ dependencies = [
     "scikit-learn",
     "vedo",
     "xdg",
+    "geopandas",
+    "shapely",
+    "pandas",
+    "scipy",
+    "plotly",
+    "alphashape",
+    "networkx",
+    "tqdm",
+
+
 ]
 
 # Command line scripts installed as part of the installation

src/py4dgeo/__init__.py

@@ -30,3 +30,5 @@
 )
 
 from py4dgeo.pbm3c2 import *
+
+from py4dgeo.levelset import LevelSetAlgorithm