Imagery: add support of original class value in sig file (OSGeo#2425)

* Imagery: use original raster values for classification output in workflow i.gensig -> i.maxlik (fixes bug OSGeo#2413)

imagery/i.gensig/get_train.c

@@ -39,6 +39,7 @@ int get_training_classes(struct files *files, struct Signature *S)
     /* convert this to an array */
     Rast_rewind_cell_stats(&cell_stats);
     n = 0;
+    S->have_oclass = 1;
     while (Rast_next_cell_stat(&cat, &count, &cell_stats)) {
 	if (count > 1) {
 	    I_new_signature(S);
@@ -48,6 +49,8 @@ int get_training_classes(struct files *files, struct Signature *S)
 		      Rast_get_c_cat(&cat, &files->training_labels),
 		      sizeof(S->sig[n].desc)
 		);
+        S->sig[n].desc[255] = '\0';  /* desc is limited to 256 */
+        S->sig[n].oclass = cat;
 	    n++;
 	}
 	else

imagery/i.gensig/testsuite/data/b1.ascii

@@ -0,0 +1,14 @@
+north: 5
+south:  0
+east:  5
+west:   0
+rows:   5
+cols:   5
+null:   n
+type:   float
+
+3.1 2.9 8.8 8.9 9.0
+2.8 2.7 8.9 9.0 9.1
+6.5 6.8 8.7 8.8 8.9
+6.9 6.7 2.8 2.9 3.1
+7.1 6.8 2.7 3.0 3.1

imagery/i.gensig/testsuite/data/b2.ascii

@@ -0,0 +1,14 @@
+north: 5
+south:  0
+east:  5
+west:   0
+rows:   5
+cols:   5
+null:   n
+type:   int
+
+8 8 1 1 2
+7 7 2 1 2
+9 9 2 2 1
+8 9 7 7 6
+8 8 8 7 8

imagery/i.gensig/testsuite/data/train.ascii

@@ -0,0 +1,14 @@
+north: 5
+south:  0
+east:  5
+west:   0
+rows:   5
+cols:   5
+null:   n
+type:   int
+
+1 n n n 6
+n 1 6 6 n
+9 n n n 6
+9 n n n 1
+n 9 n 1 n

imagery/i.gensig/testsuite/test_i_gensig.py

@@ -0,0 +1,159 @@
+"""
+Name:      i.gensig general functionality
+Purpose:   Test ability to generate signature files
+
+Author:    Maris Nartiss
+Copyright: (C) 2022 by Maris Nartiss and the GRASS Development Team
+Licence:   This program is free software under the GNU General Public
+           License (>=v2). Read the file COPYING that comes with GRASS
+           for details.
+"""
+
+import os
+import stat
+import ctypes
+import shutil
+
+from grass.pygrass import utils
+from grass.pygrass.gis import Mapset
+from grass.script.core import tempname
+
+from grass.gunittest.case import TestCase
+from grass.gunittest.main import test
+
+from grass.lib.gis import G_mapset_path
+from grass.lib.raster import Rast_write_semantic_label
+from grass.lib.imagery import (
+    Signature,
+    Ref,
+    I_init_group_ref,
+    I_add_file_to_group_ref,
+    I_put_group_ref,
+    I_put_subgroup_ref,
+    I_fopen_signature_file_old,
+    I_read_signatures,
+)
+
+
+class SuccessTest(TestCase):
+    """Test successful generation of a sig file"""
+
+    @classmethod
+    def setUpClass(cls):
+        """Ensures expected computational region and generated data"""
+        cls.use_temp_region()
+        cls.runModule("g.region", n=5, s=0, e=5, w=0, res=1)
+        cls.data_dir = "data"
+        cls.mpath = utils.decode(G_mapset_path())
+        cls.mapset_name = Mapset().name
+
+        # Load imagery bands and training set
+        cls.b1 = tempname(10)
+        cls.runModule(
+            "r.in.ascii",
+            input=os.path.join(cls.data_dir, "b1.ascii"),
+            output=cls.b1,
+        )
+        cls.b2 = tempname(10)
+        cls.runModule(
+            "r.in.ascii",
+            input=os.path.join(cls.data_dir, "b2.ascii"),
+            output=cls.b2,
+        )
+        cls.train = tempname(10)
+        cls.runModule(
+            "r.in.ascii",
+            input=os.path.join(cls.data_dir, "train.ascii"),
+            output=cls.train,
+        )
+        cls.group = tempname(10)
+        cls.semantic_label1 = "The_Doors"
+        cls.semantic_label2 = "The_Who"
+        Rast_write_semantic_label(cls.b1, cls.semantic_label1)
+        Rast_write_semantic_label(cls.b2, cls.semantic_label2)
+        Rg = Ref()
+        I_init_group_ref(ctypes.byref(Rg))
+        I_add_file_to_group_ref(cls.b1, cls.mapset_name, ctypes.byref(Rg))
+        I_add_file_to_group_ref(cls.b2, cls.mapset_name, ctypes.byref(Rg))
+        I_put_group_ref(cls.group, ctypes.byref(Rg))
+        Rs = Ref()
+        I_init_group_ref(ctypes.byref(Rs))
+        I_add_file_to_group_ref(cls.b1, cls.mapset_name, ctypes.byref(Rs))
+        I_add_file_to_group_ref(cls.b2, cls.mapset_name, ctypes.byref(Rs))
+        I_put_subgroup_ref(cls.group, cls.group, ctypes.byref(Rs))
+
+        # Location of target signature file
+        cls.sig_name1 = tempname(10)
+        cls.sig_dir1 = f"{cls.mpath}/signatures/sig/{cls.sig_name1}"
+
+    @classmethod
+    def tearDownClass(cls):
+        """Remove the temporary region and generated data"""
+        cls.del_temp_region()
+        shutil.rmtree(cls.sig_dir1, ignore_errors=True)
+        cls.runModule("g.remove", flags="f", type="raster", name=cls.b1, quiet=True)
+        cls.runModule("g.remove", flags="f", type="raster", name=cls.b2, quiet=True)
+        cls.runModule("g.remove", flags="f", type="raster", name=cls.train, quiet=True)
+
+    def test_creation(self):
+        """Test creating a signature"""
+        self.assertModule(
+            "i.gensig",
+            trainingmap=self.train,
+            group=self.group,
+            subgroup=self.group,
+            signaturefile=self.sig_name1,
+            quiet=True,
+        )
+
+        # File must be present
+        sig_stat = os.stat(f"{self.sig_dir1}/sig")
+        self.assertTrue(stat.S_ISREG(sig_stat.st_mode))
+
+        # Compare values within sig file
+        Sn = Signature()
+        p_old_sigfile = I_fopen_signature_file_old(self.sig_name1)
+        ret = I_read_signatures(p_old_sigfile, ctypes.byref(Sn))
+        self.assertEqual(ret, 1)
+        self.assertEqual(Sn.nbands, 2)
+        self.assertEqual(Sn.nsigs, 3)
+        self.assertEqual(Sn.have_oclass, 1)
+        semantic_label = utils.decode(
+            ctypes.cast(Sn.semantic_labels[0], ctypes.c_char_p).value
+        )
+        self.assertEqual(semantic_label, self.semantic_label1)
+        semantic_label = utils.decode(
+            ctypes.cast(Sn.semantic_labels[1], ctypes.c_char_p).value
+        )
+        self.assertEqual(semantic_label, self.semantic_label2)
+        # 1
+        self.assertEqual(Sn.sig[0].status, 1)
+        self.assertEqual(Sn.sig[0].have_color, 0)
+        self.assertEqual(Sn.sig[0].npoints, 4)
+        self.assertEqual(Sn.sig[0].oclass, 1)
+        self.assertTrue(abs(Sn.sig[0].mean[0] - 2.9) < 0.1)
+        self.assertTrue(abs(Sn.sig[0].mean[1] - 7.0) < 0.1)
+        self.assertTrue(abs(Sn.sig[0].var[0][0] - 0.03) < 0.01)
+        self.assertTrue(abs(Sn.sig[0].var[1][1] - 0.6) < 0.1)
+        # 6
+        self.assertEqual(Sn.sig[1].status, 1)
+        self.assertEqual(Sn.sig[1].have_color, 0)
+        self.assertEqual(Sn.sig[1].npoints, 4)
+        self.assertEqual(Sn.sig[1].oclass, 6)
+        self.assertTrue(abs(Sn.sig[1].mean[0] - 8.9) < 0.1)
+        self.assertTrue(abs(Sn.sig[1].mean[1] - 1.5) < 0.1)
+        self.assertTrue(abs(Sn.sig[1].var[0][0] - 0.003) < 0.001)
+        self.assertTrue(abs(Sn.sig[1].var[1][1] - 0.3) < 0.1)
+        # 9
+        self.assertEqual(Sn.sig[2].status, 1)
+        self.assertEqual(Sn.sig[2].have_color, 0)
+        self.assertEqual(Sn.sig[2].npoints, 3)
+        self.assertEqual(Sn.sig[2].oclass, 9)
+        self.assertTrue(abs(Sn.sig[2].mean[0] - 6.7) < 0.1)
+        self.assertTrue(abs(Sn.sig[2].mean[1] - 8.3) < 0.1)
+        self.assertTrue(abs(Sn.sig[2].var[0][0] - 0.043) < 0.01)
+        self.assertTrue(abs(Sn.sig[2].var[1][1] - 0.3) < 0.1)
+
+
+if __name__ == "__main__":
+    test()

imagery/i.maxlik/main.c

@@ -120,7 +120,13 @@ int main(int argc, char *argv[])
 	    Rast_get_d_row(cellfd[band], cell[band], row);
 
 	classify(class_cell, reject_cell, ncols);
-	Rast_put_row(class_fd, class_cell, CELL_TYPE);
+    if (S.have_oclass) {
+        for (int col = 0; col < ncols; col++) {
+            /* Predicted classes start at 1 but signature array is 0 based */
+            class_cell[col] = S.sig[class_cell[col] - 1].oclass;
+        }
+    }
+    Rast_put_row(class_fd, class_cell, CELL_TYPE);
 	if (reject_fd > 0)
 	    Rast_put_row(reject_fd, reject_cell, CELL_TYPE);
     }
@@ -133,8 +139,11 @@ int main(int argc, char *argv[])
     Rast_init_cats("Maximum Likelihood Classification", &cats);
     for (i = 0; i < S.nsigs; i++) {
 	if (*S.sig[i].desc) {
-	    cat = i + 1;
-	    Rast_set_c_cat(&cat, &cat, S.sig[i].desc, &cats);
+	    if (S.have_oclass)
+            cat = S.sig[i].oclass;
+        else
+            cat = i + 1;
+        Rast_set_c_cat(&cat, &cat, S.sig[i].desc, &cats);
 	}
     }
     Rast_write_cats(class_name, &cats);