- added image module with sub image and normalize

- improved tests

src/rbm/image.py

@@ -0,0 +1,46 @@
+from rbm.matrix import Mat, np
+
+def normalize(image: Mat):
+	image_shape = image.shape
+	mean_training_batch = np.reshape(np.repeat(np.mean(image, axis=1), image_shape[1], axis=0), image_shape)
+	var_training_batch = np.reshape(np.repeat(np.std(image, axis=1), image_shape[1], axis=0), image_shape)
+	result = (image - mean_training_batch) / var_training_batch
+	return result
+
+class SubImage:
+	def __init__(self, sx: int, sy: int, px: int, py: int):
+		self.sx = sx
+		self.sy = sy
+		self.px = px
+		self.py = py
+		self.x = 0
+		self.y = 0
+
+	def check(self, image: Mat):
+		nx = (image.shape[0] - self.sx)
+		ny = (image.shape[1] - self.sy)
+		if nx % self.px:
+			raise ValueError
+		if ny % self.py:
+			raise ValueError
+
+		nx = int(nx / self.px) + 1
+		ny = int(ny / self.py) + 1
+		return nx, ny
+
+	def __call__(self, images: Mat):
+		return self.process_image(images)
+
+	def process_image(self, image: Mat):
+		nx, ny = self.check(image[0, 0, :, :])
+		result = np.zeros((image.shape[0]*nx*ny, image.shape[1], self.sx, self.sy))
+		k = 0
+		for n in range(image.shape[0]):
+			for x in range(0, image.shape[2] - self.sx + 1, self.px):
+				for y in range(0, image.shape[3] - self.sy + 1, self.py):
+					for c in range(image.shape[1]):
+						si = image[n, c, x:x+self.sx, y:y+self.sy]
+						result[k, c] = si
+					k += 1
+
+		return result

src/tests/test_linear.py

@@ -1,6 +1,7 @@
 from rbm.matrix import Mat, np
 from rbm.entity import Entity, EntityParams, TrainingParams
 from rbm.model import Model
+from rbm.image import normalize
 
 WORK_DIR = "../../results"
 USE_OPTIMIZER = True
@@ -45,12 +46,10 @@ def linear():
 	training_batch = np.random.randn(N_CASES, N_VIS, dtype=np.float64)
 
 	# Normalize training data
-	mean_training_batch = np.reshape(np.repeat(np.mean(training_batch, axis=1), N_VIS, axis=0), training_batch.shape)
-	var_training_batch = np.reshape(np.repeat(np.std(training_batch, axis=1), N_VIS, axis=0), training_batch.shape)
-	training_batch = (training_batch - mean_training_batch) / var_training_batch
+	training_batch_norm = normalize(training_batch)
 
 	# Train layer
-	model.train(training_batch)
+	model.train(training_batch_norm)
 
 	# Save weights
 	model.save()

src/tests/test_sub_image.py

@@ -0,0 +1,19 @@
+from rbm.image import SubImage
+from rbm.matrix import np
+
+
+if __name__ == "__main__":
+	n = 2
+	c = 3
+	w = 8
+	h = 8
+	img = np.reshape(np.linspace(1, n*c*w*h, num=n*c*w*h, dtype=np.float64), (n, c, w, h))
+#	print(img)
+
+	sub = SubImage(4, 4, 1, 1)
+
+	images = sub(img)
+	print(images)
+
+
+	print("Test: [passed]")