Merge pull request #130 from nocthulhu/add-Numerecizer-files

Added Numerecizer files

Author: canbula

Projects/Numerecizer/calibration/__init__.py

@@ -0,0 +1,3 @@
+from .calibration import Calibration
+
+__all__ = ['Calibration']

Projects/Numerecizer/calibration/__pycache__/__init__.cpython-312.pyc

@@ -0,0 +1,265 @@
+import cv2
+import numpy as np
+from PyQt5.QtCore import QPointF
+from PyQt5.QtWidgets import QDialog
+from point import Point
+from ui.calibration_dialog import CalibrationDialog
+import random
+class Calibration:
+    """Class to manage calibration of images to real-world coordinates."""
+
+    def __init__(self, main_window):
+        self.main_window = main_window
+        self.calibration_points = []
+        self.transformation_matrix = None
+        self.inverse_transformation_matrix = None
+        self.calibration_done = False
+        self.automatic_calibration_mode = False
+        self.calibration_cancelled = False
+
+    def add_calibration_point(self, point: QPointF, automatic=False):
+        """Adds a calibration point and triggers dialog for real coordinates input."""
+        if len(self.calibration_points) < 4:  # Allow selecting 4 calibration points
+            point_obj = Point(point, point_type='calibration')
+            self.calibration_points.append(point_obj)
+
+            self.main_window.image_view.highlight_point(point_obj)
+            self.main_window.image_view.update_scene()
+            dialog = CalibrationDialog(self.main_window, automatic_calibration=automatic)
+            dialog.setWindowTitle(f"Enter Real Coordinates for Point {len(self.calibration_points)}")
+            result = dialog.exec()
+            if result == QDialog.Accepted:
+                real_x, real_y = dialog.real_coordinates
+                point_obj.set_real_coordinates(QPointF(real_x, real_y))
+                self.main_window.image_view.delete_highlight(point_obj)
+                self.main_window.image_view.draw_calibration_points(self.calibration_points)
+                if len(self.calibration_points) == 4:  # Calculate, refine transformation matrix
+                    self.calculate_transformation_matrix()
+                    self.refine_calibration()
+            elif result == 1000:  # Next point
+                self.calibration_points.pop()
+                self.main_window.image_view.delete_highlight(point_obj)
+                new_point = self.select_random_point()
+                if new_point is not None and not self.calibration_cancelled:
+                    self.add_calibration_point(new_point, automatic=True)
+            else:
+                self.calibration_points.pop()
+                self.main_window.image_view.delete_highlight(point_obj)
+                if automatic:
+                    self.calibration_cancelled = True  # Mark calibration as cancelled
+                    self.clear_calibration_points()
+
+    def clear_calibration_points(self):
+        """Clears all calibration points."""
+        self.main_window.image_view.clear_calibration_points()
+        self.calibration_points = []
+        self.calibration_done = False
+
+    def calculate_transformation_matrix(self):
+        """Calculates the transformation matrix based on calibration points."""
+        if len(self.calibration_points) != 4:
+            raise ValueError("Exactly 4 calibration points are required to calculate the transformation matrix.")
+
+        image_points = np.array(
+            [[p.get_image_coordinates().x(), p.get_image_coordinates().y()] for p in self.calibration_points],
+            dtype=np.float32)
+        real_coords = np.array(
+            [[p.get_real_coordinates().x(), p.get_real_coordinates().y()] for p in self.calibration_points],
+            dtype=np.float32)
+
+        self.transformation_matrix, _ = cv2.findHomography(image_points, real_coords)
+        self.inverse_transformation_matrix = np.linalg.inv(self.transformation_matrix)
+
+        self.calibration_done = True
+        self.main_window.interpolationAction.setEnabled(True)
+
+    def refine_calibration(self, iterations=500, termination_eps=1e-6):
+        """Refines the calibration matrix using iterative optimization."""
+        if len(self.calibration_points) != 4:
+            raise ValueError("Exactly 4 calibration points are required to refine the calibration matrix.")
+
+        image_points = np.array(
+            [[p.get_image_coordinates().x(), p.get_image_coordinates().y()] for p in self.calibration_points],
+            dtype=np.float32)
+        real_coords = np.array(
+            [[p.get_real_coordinates().x(), p.get_real_coordinates().y()] for p in self.calibration_points],
+            dtype=np.float32)
+
+        self.transformation_matrix, _ = cv2.findHomography(image_points, real_coords, method=cv2.RANSAC, ransacReprojThreshold=5.0, maxIters=iterations, confidence=0.99)
+        self.inverse_transformation_matrix = np.linalg.inv(self.transformation_matrix)
+
+        self.calibration_done = True
+        self.main_window.interpolationAction.setEnabled(True)
+
+    def transform_points(self, data_points):
+        """Transforms data points using the calibration matrix."""
+        if not self.calibration_done:
+            raise RuntimeError("Calibration is not complete. Please complete calibration before transforming points.")
+
+        transformed_points = []
+        for point in data_points:
+            img_coords = np.array([[point.get_image_coordinates().x(), point.get_image_coordinates().y(), 1.0]],
+                                  dtype=np.float32)
+            real_coords = cv2.perspectiveTransform(np.array([img_coords]), self.transformation_matrix)
+            point.set_real_coordinates(QPointF(real_coords[0][0][0], real_coords[0][0][1]))
+            transformed_points.append(point)
+
+        return transformed_points
+
+    def inverse_transform_point(self, x, y):
+        """Transforms real-world coordinates back to image coordinates using the inverse calibration matrix."""
+        if self.inverse_transformation_matrix is None:
+            raise ValueError("Inverse transformation matrix has not been calculated yet.")
+
+        real_coords = np.array([[x, y]], dtype=np.float32).reshape(-1, 1, 2)
+        img_coords = cv2.perspectiveTransform(real_coords, self.inverse_transformation_matrix)
+
+        return QPointF(img_coords[0][0][0], img_coords[0][0][1])
+
+    def image_to_real_coordinates(self, point):
+        """Transforms image coordinates to real-world coordinates using the calibration matrix."""
+        if not self.calibration_done:
+            raise RuntimeError("Calibration is not complete. Please complete calibration before transforming points.")
+
+        img_coords = np.array([[point.x(), point.y()]], dtype=np.float32).reshape(-1, 1, 2)
+        real_coords = cv2.perspectiveTransform(img_coords, self.transformation_matrix)
+        return QPointF(real_coords[0][0][0], real_coords[0][0][1])
+
+    def refine_calibration(self, iterations=500, termination_eps=1e-6):
+        """Refines the calibration matrix using iterative optimization."""
+        if len(self.calibration_points) != 4:
+            raise ValueError("Exactly 4 calibration points are required to refine the calibration matrix.")
+
+        image_points = np.array(
+            [[p.get_image_coordinates().x(), p.get_image_coordinates().y()] for p in self.calibration_points],
+            dtype=np.float32)
+        real_coords = np.array(
+            [[p.get_real_coordinates().x(), p.get_real_coordinates().y()] for p in self.calibration_points],
+            dtype=np.float32)
+
+        # Use cv2.findHomography with RANSAC to refine the transformation matrix
+        self.transformation_matrix, _ = cv2.findHomography(image_points, real_coords, method=cv2.RANSAC, ransacReprojThreshold=5.0, maxIters=iterations, confidence=0.99)
+        self.inverse_transformation_matrix = np.linalg.inv(self.transformation_matrix)
+
+        self.calibration_done = True
+        self.main_window.interpolationAction.setEnabled(True)
+    def advanced_corner_detection(self, image):
+        """Improves corner detection using optimized algorithms."""
+        gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+        gray = cv2.GaussianBlur(gray, (5, 5), 0)
+
+        # Enhanced contrast and adaptive thresholding
+        clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
+        gray = clahe.apply(gray)
+        gray = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 11, 2)
+
+        # Edge detection
+        edges = cv2.Canny(gray, 50, 150, apertureSize=3)
+
+        # Line detection using Hough Transform
+        lines = cv2.HoughLinesP(edges, 1, np.pi / 180, 50, minLineLength=50, maxLineGap=10)
+
+        line_img = np.zeros_like(image)
+        if lines is not None:
+            for line in lines:
+                x1, y1, x2, y2 = line[0]
+                cv2.line(line_img, (x1, y1), (x2, y2), (255, 0, 0), 2)
+
+        # Find line intersections
+        intersections = self.find_intersections(lines)
+
+        # Corner detection using Shi-Tomasi algorithm on the edge image
+        corners = cv2.goodFeaturesToTrack(edges, maxCorners=100, qualityLevel=0.01, minDistance=10)
+        corners = np.float32(corners)
+
+        # Refining corner locations using cornerSubPix for sub-pixel accuracy
+        criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 30, 0.01)
+        corners = cv2.cornerSubPix(gray, corners, (5, 5), (-1, -1), criteria)
+
+        # Only keep corners that are close to detected lines and intersections
+        refined_corners = []
+        for corner in corners:
+            x, y = corner.ravel()
+            if self.is_near_line(x, y, lines) or self.is_near_intersection(x, y, intersections):
+                refined_corners.append(QPointF(x, y))
+                self.main_window.image_view.draw_detected_corners(refined_corners)
+
+        self.main_window.image_view.set_image(image)
+        self.main_window.update_image()
+
+        return refined_corners
+
+    def is_near_line(self, x, y, lines, threshold=5):
+        """Checks if a point (x, y) is near any of the detected lines."""
+        if lines is not None:
+            for line in lines:
+                x1, y1, x2, y2 = line[0]
+                distance = self.point_line_distance(x, y, x1, y1, x2, y2)
+                if distance < threshold:
+                    return True
+        return False
+
+    def point_line_distance(self, px, py, x1, y1, x2, y2):
+        """Calculates the minimum distance from a point to a line segment."""
+        line_mag = np.sqrt((x2 - x1) ** 2 + (y2 - y1) ** 2)
+        if line_mag < 1e-10:
+            return np.sqrt((px - x1) ** 2 + (py - y1) ** 2)
+        u = ((px - x1) * (x2 - x1) + (py - y1)) / line_mag ** 2
+        if u < 0 or u > 1:
+            ix = min(max(x1, x2), max(px, px))
+            iy = min(max(y1, y2), max(py, py))
+        else:
+            ix = x1 + u * (x2 - x1)
+            iy = y1 + u * (y2 - y1)
+        return np.sqrt((px - ix) ** 2 + (py - iy) ** 2)
+
+    def find_intersections(self, lines):
+        """Finds intersections between lines."""
+        intersections = []
+        if lines is not None:
+            for i in range(len(lines)):
+                for j in range(i + 1, len(lines)):
+                    x1, y1, x2, y2 = lines[i][0]
+                    x3, y3, x4, y4 = lines[j][0]
+                    denom = (x1 - x2) * (y3 - y4) - (y1 - y2) * (x3 - x4)
+                    if denom != 0:
+                        intersect_x = ((x1 * y2 - y1 * x2) * (x3 - x4) - (x1 - x2) * (x3 * y4 - y3 * x4)) / denom
+                        intersect_y = ((x1 * y2 - y1 * x2) * (y3 - y4) - (y1 - y2) * (x3 * y4 - y3 * x4)) / denom
+                        intersections.append((intersect_x, intersect_y))
+        return intersections
+
+    def is_near_intersection(self, x, y, intersections, threshold=5):
+        """Checks if a point (x, y) is near any of the detected intersections."""
+        for intersect_x, intersect_y in intersections:
+            distance = np.sqrt((x - intersect_x) ** 2 + (y - intersect_y) ** 2)
+            if distance < threshold:
+                return True
+        return False
+
+    def automatic_calibration(self, image):
+        """Automatically calibrates the image using enhanced corner detection."""
+        print("Starting automatic calibration...")
+        self.automatic_calibration_mode = True
+        self.calibration_cancelled = False  # Reset cancellation flag
+        corners = self.advanced_corner_detection(image)
+
+        if len(corners) >= 10:
+            corners = random.sample(corners, 4)  # Take 4 random corners
+            self.calibration_points = []
+            for corner in corners:
+                if not self.calibration_cancelled:
+                    self.add_calibration_point(corner, automatic=True)
+            if len(self.calibration_points) == 4:
+                self.calculate_transformation_matrix()
+            else:
+                self.clear_calibration_points()
+        else:
+            print("Not enough corners detected for calibration.")
+        self.automatic_calibration_mode = False
+
+    def select_random_point(self):
+        """Selects a random point from detected corners for automatic calibration."""
+        corners = self.advanced_corner_detection(self.main_window.image_processor.image)
+        if corners:
+            return random.choice(corners)
+        return None

Projects/Numerecizer/calibration/__pycache__/calibration.cpython-312.pyc

@@ -0,0 +1,3 @@
+from .extraction import DataExtraction
+
+__all__ = ['DataExtraction']

Projects/Numerecizer/calibration/calibration.py

@@ -0,0 +1,66 @@
+import cv2
+from PyQt5.QtCore import QPointF
+from point import Point
+import numpy as np
+
+class DataExtraction:
+    def __init__(self, calibration, main_window):
+        self.data_points = []
+        self.temp_points = []  # points automatic extraction
+        self.calibration = calibration
+        self.main_window = main_window
+
+    def add_data_point(self, scene_pos):
+        """Add a data point at the given scene position."""
+        real_coordinates = self.calibration.image_to_real_coordinates(scene_pos)
+        point = Point(scene_pos, real_coordinates, point_type='data')
+        self.data_points.append(point)
+        self.main_window.image_view.draw_data_points(self.data_points)
+        if self.main_window.interpolation_mode:
+            self.main_window.interpolation.interpolate_data(self.data_points)
+        self.main_window.show_data_points()
+
+    def delete_data_point(self, index):
+        """Deletes a data point at the given index."""
+        if 0 <= index < len(self.data_points):
+            del self.data_points[index]
+            self.main_window.image_view.draw_data_points(self.data_points)
+            if self.main_window.interpolation_mode:
+                self.main_window.interpolation.interpolate_data(self.data_points)
+            self.main_window.show_data_points()
+
+    def get_data_points(self):
+        """Returns the list of data points."""
+        return self.data_points
+
+    def clear_temp_points(self):
+        """Clears the temporary points found during automatic extraction."""
+        self.temp_points = []
+    def clear_data_points(self):
+        """Clears the data points"""
+        self.data_points = []
+    def automatic_extraction(self, image):
+        """Automatically detects data points from the image for visualization."""
+        gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+        blurred = cv2.GaussianBlur(gray, (5, 5), 0)
+
+        # Enhanced contrast and adaptive thresholding
+        clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
+        gray = clahe.apply(blurred)
+        threshold = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 11, 2)
+
+        # Finding contours
+        contours, _ = cv2.findContours(threshold, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+
+        self.clear_temp_points()
+        for contour in contours:
+            if cv2.contourArea(contour) > 10:  # Minimum area to filter out noise
+                M = cv2.moments(contour)
+                if M["m00"] != 0:
+                    cX = int(M["m10"] / M["m00"])
+                    cY = int(M["m01"] / M["m00"])
+                    point = QPointF(cX, cY)
+                    self.temp_points.append(point)
+
+        self.main_window.image_view.draw_detected_points(self.temp_points)
+        return image

Projects/Numerecizer/data_extraction/__init__.py

@@ -0,0 +1,3 @@
+from .data_exporter import DataExporter
+
+__all__ = ['DataExporter']