compute_traps_mask now returns a traps mask, as expected

- to do this I had to split test_template_matcher into 2 unit tests (one of them being  a unit test just for template matching)

src/lipase/catalog.py

@@ -132,6 +132,8 @@ class Sequence(object):
         """ returns a subset of the sequence as an hyperstack
 
         :param list(str) selected_channel_ids:
+        :param list(int) selected_frames:
+        :param list(int) selected_slices:
         :return IHyperStack: the resulting hyperstack
         """
         if selected_frames is None:

src/lipase/imageengine.py

@@ -40,7 +40,10 @@ class PixelType:
 class IImage(ABC):
 
     @abc.abstractmethod
-    def clone(self):
+    def clone(self, clone_pixel_type=None):
+        """
+        :param int clone_pixel_type: PixelType enum. If None, then the clone's pixel type is the same as the pixel type of the source image
+        """
         pass
 
     @abc.abstractmethod
@@ -80,6 +83,12 @@ class IImage(ABC):
         """
         pass
     
+    @abc.abstractmethod
+    def get_mean_value(self):
+        """
+        :rtype float: the mean pixel value in the image
+        """
+        pass
 
 class IHyperStack(ABC):
 

src/lipase/imagej/ijimageengine.py

@@ -43,9 +43,18 @@ class IJImage(IImage):
             stack_name = ''
             self.ij_image = IJ.createHyperStack(stack_name, width, height, 1, 1, 1, PixelType.get_num_bits(pixel_type))
 
-    def clone(self):
+    def clone(self, clone_pixel_type=None):
-        copy = IJImage(self.image_engine, width=self.width(), height=self.height(), pixel_type=self.get_pixel_type())
+        if clone_pixel_type == None:
-        copy.ij_image.setProcessor( self.ij_image.getProcessor().duplicate() )
+            clone_pixel_type = self.get_pixel_type()
+        copy = IJImage(self.image_engine, width=self.width(), height=self.height(), pixel_type=clone_pixel_type)
+        src_processor = self.ij_image.getProcessor()
+        clone_processor = { 
+            PixelType.U8: src_processor.convertToByteProcessor(),
+            PixelType.U16: src_processor.convertToShortProcessor(),
+            PixelType.F32: src_processor.convertToFloatProcessor(),
+             }[clone_pixel_type]
+        copy.ij_image.setProcessor( clone_processor )
+        assert copy.get_pixel_type() == clone_pixel_type
         return copy
 
     def width(self):
@@ -78,11 +87,15 @@ class IJImage(IImage):
         processor = self.ij_image.getProcessor()
         # processor.getMin() and processor.getMax() give a higher min and a lower max, probably because they only consider a subpart of the image (this is badly documented)
         # so we use processor.getStats instead
-        image_stats = processor.getStats()  # :type ImageStatistics is:
+        image_stats = processor.getStats()  # :type ImageStatistics image_stats:
         min_value = image_stats.min
         max_value = image_stats.max
         return (min_value, max_value)
 
+    def get_mean_value(self):
+        processor = self.ij_image.getProcessor()
+        image_stats = processor.getStats()  # :type ImageStatistics image_stats:
+        return image_stats.mean
 
 class IJHyperStack(IHyperStack):
 
@@ -303,7 +316,13 @@ class IJImageEngine(IImageEngine):
         return image  # pylint: disable=unreachable
 
     def match_template(self, src_image, template_image):
-
+        cv_match_template_supported_pixel_types = [PixelType.U8, PixelType.F32]
+        if src_image.get_pixel_type() not in cv_match_template_supported_pixel_types:
+            print('converting src_image')
+            src_image = src_image.clone(PixelType.F32)
+        if template_image.get_pixel_type() not in cv_match_template_supported_pixel_types:
+            print('converting template_image')
+            template_image = template_image.clone(PixelType.F32)
         # import org.opencv.imgproc.Imgproc;
         #
         # /*

src/lipase/traps_detector.py

@@ -116,5 +116,5 @@ class TrapsDetector(object):
         #non_uniform_stack = sequence.as_stack()
         #uniform_stack = IImageEngine.get_instance().divide(non_uniform_stack, white_estimate)
         # IImageEngine.get_instance().save_as_tiff(white_estimate, './white_estimate.tiff')
-        return matches
+        return traps_mask
         

tests/test0001.py

@@ -45,6 +45,30 @@ class TestLipase(unittest.TestCase):
         non_uniform_sequence = self.catalog.sequences['res_soleil2018/GGH/GGH_2018_cin2_phiG_I_327_vis_-40_1/Pos0']
         uniform_sequence = correct_non_uniform_lighting(non_uniform_sequence, 'DM300_nofilter_vis', white_estimator=WhiteEstimator(open_size=75, close_size=75, average_size=75))
 
+    def test_template_matcher(self):
+        sequence = self.catalog.sequences['res_soleil2018/GGH/GGH_2018_cin2_phiG_I_327_vis_-40_1/Pos0']
+        stack = sequence.as_hyperstack(['DM300_nofilter_vis'], selected_frames=[0])
+        first_image = stack.get_image(frame_index=0)
+        x_min = 423
+        x_max = 553
+        y_min = 419
+        y_max = 533
+        template_trap_aabb = Aabb(x_min, y_min, x_max, y_max)
+        template_trap_image = first_image.get_subimage(template_trap_aabb)
+        for image in [first_image, template_trap_image]:
+            print(image.get_pixel_type(), image.width(), image.height())
+        # the typical value of peaks is -2.e10 and the value between peaks is below -8.0e10
+        threshold = -3.0e10
+        tolerance = 1.0e10
+        maxima_finder = MaximaFinder(threshold, tolerance)
+        template_matcher = TemplateMatcher(maxima_finder)
+        matches = template_matcher.match_template(first_image, template_trap_image)
+        num_traps = len(matches)
+        print("number of traps found : %d" % num_traps)
+        num_expected_traps = 13  # 13 traps are completely visible in the first image
+        self.assertAlmostEqual(len(matches), num_expected_traps, delta=1.0)
+
+
     def test_traps_detector(self):
         # the typical value of peaks is -500 and the value between peaks is below -2500
         threshold = -1500.0
@@ -58,12 +82,14 @@ class TestLipase(unittest.TestCase):
         y_min = 419
         y_max = 533
         trap_aabb = Aabb(x_min, y_min, x_max, y_max)
-        matches = traps_detector.compute_traps_mask(sequence, 'DM300_nofilter_vis', trap_aabb)
+        traps_mask = traps_detector.compute_traps_mask(sequence, 'DM300_nofilter_vis', trap_aabb)
-
+        measured_mean_value = traps_mask.get_mean_value()
-        num_traps = len(matches)
+        expected_traps_coverage = 0.07909
-        print("number of traps found : %d" % num_traps)
+        traps_pixel_value = 255.0
-        num_expected_traps = 13  # 13 traps are completely visible in the 
+        expected_mean_value = expected_traps_coverage * traps_pixel_value
-        self.assertAlmostEqual(len(matches), num_expected_traps, delta=1.0)
+        print("expected_mean_value: %f" % expected_mean_value)
+        print("measured_mean_value: %f" % measured_mean_value)
+        self.assertAlmostEqual(measured_mean_value, expected_mean_value, delta=0.01)
 
 def run_script():