add patch evluation

.gitignore

@@ -9,4 +9,5 @@ orisys.egg-info/
 /dist_single_pyd/
 /old_orisys/
 /dist_nuitka/
-/build_nuitka/
+/build_nuitka
+/data/

example1_basic_pipeline.py

@@ -19,7 +19,8 @@ def main():
     # 步骤1：创建传感器对象 (必要步骤)，config_name为配置文件名（内置配置文件名）或路径（自定义配置文件路径）
 
     # 方式1: 从视频文件读取
-    sensor = orisys.Sensor("./s1.mp4", isstitch=True, cuda=True, verbose=True,config_name="prototype")
+    sensor = orisys.Sensor("./ov2710_uv_2.mp4", isstitch=True, cuda=True, verbose=True,config_name="./config/lens_test.json")
+
     
     # 方式2: 从摄像头读取（取消注释使用） 
     # sensor = orisys.Sensor(0, cuda=True, verbose=True, config_name="prototype")
@@ -31,7 +32,7 @@ def main():
         # 步骤2：信息提取
         # 步骤 2.1: 获取并拼接图像 (必要步骤)
         sensor.get_img() 
-        
+        print("img_size_raw:", sensor.img_size_raw)
         # 步骤 2.2: 计算形变场 (必要步骤)
         sensor.compute_deformation()
         # 步骤 2.3: 计算触碰点 

patch_generation/patch_generation.py

@@ -0,0 +1,53 @@
+#%%
+import numpy as np
+import cv2
+
+def generate_color_pattern(h, w, d, r):
+    # Initialize the pattern with zeros
+    pattern = np.zeros((h, w, 3))
+
+    # Calculate the number of patches in each dimension
+    num_patches_x = h // d
+    num_patches_y = w // d
+
+    # Maximum number of attempts to find a suitable color
+    max_attempts = 1000
+
+    # Iterate over the patches
+    for i in range(num_patches_x):
+        for j in range(num_patches_y):
+            # Get the neighbors
+            neighbors = []
+            if i > 0:
+                neighbors.append(pattern[(i-1)*d:i*d, j*d:(j+1)*d])
+            if j > 0:
+                neighbors.append(pattern[i*d:(i+1)*d, (j-1)*d:j*d])
+
+            # Generate a new color
+            for attempt in range(max_attempts):
+                new_color = np.random.rand(3)
+                if all(np.min((new_color - neighbor.mean(axis=(0, 1)))**2) >= r for neighbor in neighbors):
+                    break
+            if attempt == max_attempts - 1:
+                print(f"Warning: Could not find a suitable color for patch ({i}, {j}) after {max_attempts} attempts")
+
+            # Apply the new color to the patch
+            pattern[i*d:(i+1)*d, j*d:(j+1)*d] = new_color
+
+            print('Patch ({}, {})'.format(i, j))
+
+    return pattern
+
+# Test the function
+pattern = generate_color_pattern(162, 162, 3, 0.75)
+
+# Convert the pattern to a format suitable for display with OpenCV
+pattern_display = (pattern * 255).astype(np.uint8)
+
+# # Display the pattern
+# cv2.imshow('Pattern', pattern_display)
+# cv2.waitKey(0)
+# cv2.destroyAllWindows()
+
+# #%%
+cv2.imwrite('color_pattern.png', pattern_display)

patch_generation/patch_generation_3.py

@@ -0,0 +1,139 @@
+#%%
+import numpy as np
+import cv2
+
+def merge_intervals(intervals):
+    # Merge overlapping intervals.
+    if not intervals:
+        return []
+    intervals.sort(key=lambda x: x[0])
+    merged = [intervals[0]]
+    for current in intervals[1:]:
+        last = merged[-1]
+        if current[0] <= last[1]:
+            merged[-1] = (last[0], max(last[1], current[1]))
+        else:
+            merged.append(current)
+    return merged
+
+def get_allowed_intervals(neighbor_values, delta):
+    # Compute forbidden intervals for this channel from neighbor values.
+    forbidden = []
+    for c in neighbor_values:
+        low = max(0, c - delta)
+        high = min(1, c + delta)
+        forbidden.append((low, high))
+    # Merge overlapping forbidden intervals.
+    merged = merge_intervals(forbidden)
+    allowed = []
+    # Allowed region before the first forbidden interval.
+    if merged:
+        if merged[0][0] > 0:
+            allowed.append((0, merged[0][0]))
+        # Allowed regions between forbidden intervals.
+        for i in range(1, len(merged)):
+            allowed.append((merged[i-1][1], merged[i][0]))
+        # Allowed region after the last forbidden interval.
+        if merged[-1][1] < 1:
+            allowed.append((merged[-1][1], 1))
+    else:
+        allowed.append((0, 1))
+    return allowed
+
+def sample_from_allowed(allowed):
+    # Choose a random value from the allowed intervals with probability proportional to interval length.
+    lengths = [end - start for start, end in allowed]
+    total_length = sum(lengths)
+    if total_length <= 0:
+        return None
+    point = np.random.rand() * total_length
+    cumulative = 0
+    for (start, end), length in zip(allowed, lengths):
+        cumulative += length
+        if point <= cumulative:
+            return np.random.uniform(start, end)
+    return allowed[-1][0]
+
+def sample_channel(neighbor_values, delta):
+    # Compute allowed intervals and sample a value.
+    allowed = get_allowed_intervals(neighbor_values, delta)
+    if not allowed:
+        return None
+    return sample_from_allowed(allowed)
+
+def generate_valid_color(neighbors, r):
+    """
+    Generate a random color that satisfies, for each neighbor,
+    |R - R_neighbor| >= sqrt(r), and similarly for G and B.
+    neighbors: list of neighbor colors (each is an array of shape (3,))
+    r: threshold for squared difference.
+    """
+    delta = np.sqrt(r)
+    color = np.zeros(3)
+    for channel in range(3):
+        neighbor_channel = [n[channel] for n in neighbors]
+        sampled = sample_channel(neighbor_channel, delta)
+        # If the allowed intervals are empty, use a fallback.
+        if sampled is None:
+            sampled = np.random.rand()
+        color[channel] = sampled
+    return color
+
+def generate_color_pattern(h, w, d, r):
+    """
+    Generate a dense color pattern.
+    h, w: height and width of the output image in pixels.
+    d: patch size in pixels (each patch is d x d).
+    r: randomness regulation factor. For each neighbor the condition is:
+       (|channel_new - channel_neighbor|)^2 >= r.
+    """
+    pattern = np.zeros((h, w, 3), dtype=np.float32)
+    num_patches_x = h // d
+    num_patches_y = w // d
+
+    for i in range(num_patches_x):
+        for j in range(num_patches_y):
+            # Collect neighbor patch colors (using mean because patches have uniform color).
+            neighbors = []
+            if i > 0:
+                patch_above = pattern[(i-1)*d:i*d, j*d:(j+1)*d]
+                neighbors.append(np.mean(patch_above, axis=(0, 1)))
+            if j > 0:
+                patch_left = pattern[i*d:(i+1)*d, (j-1)*d:j*d]
+                neighbors.append(np.mean(patch_left, axis=(0, 1)))
+                
+            if len(neighbors) == 0:
+                # For the first patch use a random color.
+                new_color = np.random.rand(3)
+            else:
+                new_color = generate_valid_color(neighbors, r)
+                
+            pattern[i*d:(i+1)*d, j*d:(j+1)*d] = new_color
+            print('Patch ({}, {})'.format(i, j))
+    return pattern
+
+def main():
+    # Parameters for the pattern.
+    # For example, using 100x100 pixels image with patches of 3x3 pixels.
+    height = 200
+    width = 200
+    patch_size = 3
+    # Set r to a value that gives sufficient difference.
+    # Note: r is the threshold for squared difference per channel, and the condition is
+    # |c - c_neighbor| >= sqrt(r). Very high r may make it impossible to satisfy.
+    r = 0.5
+
+    pattern = generate_color_pattern(height, width, patch_size, r)
+    # Scale to 0-255 for display and saving.
+    pattern_display = (pattern * 255).astype(np.uint8)
+
+    # Display the pattern with OpenCV.
+    cv2.imshow('Pattern', pattern_display)
+    cv2.waitKey(0)
+    cv2.destroyAllWindows()
+
+    # Save the pattern as an image file.
+    cv2.imwrite('color_pattern.png', pattern_display)
+
+if __name__ == '__main__':
+    main()

patch_generation/test_hue_gray.py

@@ -0,0 +1,224 @@
+import cv2
+import numpy as np
+import matplotlib.pyplot as plt
+def calculate_local_contrast(image_input, window_size=5):
+    """
+    Calculate Global Average Local Contrast (for fine texture detail evaluation in optical flow).
+    Args:
+        image_input: str (image file path) or numpy array (grayscale/RGB image).
+        window_size: int (odd number, default=5) — sliding window size for fine texture analysis.
+    Returns:
+        float: Global average local contrast (higher = better fine texture detail; target ≥0.2 for optical flow).
+    Raises:
+        ValueError: If image is invalid, window size is even/too small, or processing fails.
+    """
+    # ---------------------- Step 1: Input Validation & Preprocessing ----------------------
+    # Validate window size (must be odd, ≥3 to capture meaningful texture)
+    if window_size % 2 == 0 or window_size < 3:
+        raise ValueError(f"Window size must be an odd integer ≥3 (got {window_size}). Use 3/5/7.")
+    
+    # Load/convert image to grayscale (consistent with optical flow workflow)
+    if isinstance(image_input, str):
+        img = cv2.imread(image_input)
+        if img is None:
+            raise ValueError(f"Failed to load image from path: {image_input}")
+    elif isinstance(image_input, np.ndarray):
+        img = image_input.copy()
+    else:
+        raise ValueError("Input must be an image file path (str) or numpy array (image).")
+    
+    # Convert to grayscale (handle RGB/BGR)
+    if len(img.shape) == 3:
+        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
+    else:
+        gray = img  # Already grayscale
+    
+    # Normalize to 0-255 (avoid non-standard pixel ranges, e.g., float32)
+    gray = cv2.normalize(gray, None, 0, 255, cv2.NORM_MINMAX).astype(np.uint8)
+
+    # ---------------------- Step 2: Sliding Window Local Contrast Calculation ----------------------
+    # Define padding (replicate edge pixels to avoid boundary distortion)
+    pad_size = window_size // 2
+    gray_padded = cv2.copyMakeBorder(
+        gray, 
+        top=pad_size, bottom=pad_size, left=pad_size, right=pad_size,
+        borderType=cv2.BORDER_REPLICATE
+    )
+    
+    # Initialize array to store local contrast for each window
+    local_contrast_vals = []
+    
+    # Iterate over every pixel (center of the sliding window)
+    for y in range(pad_size, gray_padded.shape[0] - pad_size):
+        for x in range(pad_size, gray_padded.shape[1] - pad_size):
+            # Extract 5x5 (or custom) window around (x,y)
+            window = gray_padded[y-pad_size : y+pad_size+1, x-pad_size : x+pad_size+1]
+            
+            # Compute Michelson contrast for the window (avoid division by zero)
+            window_max = np.max(window)
+            window_min = np.min(window)
+            if window_max + window_min == 0:  # All pixels are 0 (black window)
+                lc = 0.0
+            else:
+                lc = (window_max - window_min) / (window_max + window_min)
+            
+            local_contrast_vals.append(lc)
+
+    # ---------------------- Step 3: Global Average Local Contrast ----------------------
+    global_local_contrast = np.mean(local_contrast_vals) if local_contrast_vals else 0.0
+
+    # Round for readability (4 decimal places)
+    return global_local_contrast
+
+def calculate_edge_strength_density(image_input):
+    """
+    Calculate Edge Strength and Edge Density for optical flow evaluation.
+    Args:
+        image_input: Can be (1) str (image file path) or (2) numpy array (grayscale/RGB image).
+    Returns:
+        dict: Contains two metrics:
+            - edge_strength: Average gradient magnitude of valid edges (higher = sharper edges).
+            - edge_density: Percentage of valid edge pixels (higher = more texture/edges).
+    Raises:
+        ValueError: If image cannot be loaded/processed.
+    """
+    # ---------------------- Step 1: Load/Preprocess Image ----------------------
+
+    img = image_input
+    # Convert to grayscale (handle RGB/BGR)
+    if len(img.shape) == 3:
+        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
+    else:
+        gray = img  # Already grayscale
+    
+    # Normalize to 0-255 (avoid float/non-standard pixel ranges)
+    gray = cv2.normalize(gray, None, 0, 255, cv2.NORM_MINMAX).astype(np.uint8)
+
+    # ---------------------- Step 2: Compute Gradient Magnitude (Sobel) ----------------------
+    # Sobel kernel size = 3 (balance between noise reduction and edge sharpness for optical flow)
+    sobel_x = cv2.Sobel(gray, cv2.CV_64F, 1, 0, ksize=3)  # Horizontal gradient
+    sobel_y = cv2.Sobel(gray, cv2.CV_64F, 0, 1, ksize=3)  # Vertical gradient
+    
+    # Compute gradient magnitude (edge strength raw values)
+    gradient_magnitude = np.sqrt(sobel_x**2 + sobel_y**2)
+
+    # ---------------------- Step 3: Adaptive Threshold for Valid Edges ----------------------
+    # Use median gradient to filter noise (robust to outliers)
+    median_gradient = np.median(gradient_magnitude)
+    # Threshold = 1.5× median (filters weak noise, keeps meaningful edges for optical flow)
+    edge_threshold = 1.5 * median_gradient
+    # Create binary mask for valid edges (1 = valid edge, 0 = non-edge)
+    valid_edge_mask = (gradient_magnitude > edge_threshold).astype(np.uint8)
+
+    # ---------------------- Step 4: Calculate Metrics ----------------------
+    # Edge Strength: Average gradient magnitude of valid edges
+    valid_gradient_values = gradient_magnitude[valid_edge_mask == 1]
+    edge_strength = np.mean(valid_gradient_values) if len(valid_gradient_values) > 0 else 0.0
+    
+    # Edge Density: Percentage of valid edge pixels (total pixels = height × width)
+    total_pixels = gray.shape[0] * gray.shape[1]
+    valid_edge_pixels = np.sum(valid_edge_mask)
+    edge_density = (valid_edge_pixels / total_pixels) * 100 if total_pixels > 0 else 0.0
+    # print(f"edge_strength: {edge_strength}, edge_density: {edge_density}")
+
+    # Round for readability (3 decimal places)
+    return edge_strength, edge_density
+
+def custom_gray_conversion(img, r_weight=0.299, g_weight=0.587, b_weight=0.114, 
+                           red_boost=1.2, green_reduce=0.8):
+    """
+    Custom grayscale calculation with tunable RGB weights + red/green contrast enhancement
+    Args:
+        image_path: Path to input image
+        r_weight: Weight for Red channel (default: 0.299)
+        g_weight: Weight for Green channel (default: 0.587)
+        b_weight: Weight for Blue channel (default: 0.114)
+        red_boost: Multiplier to boost red region grayscale (>1 = brighter)
+        green_reduce: Multiplier to reduce green region grayscale (<1 = darker)
+    Returns:
+        default_gray: Grayscale with standard weights
+        custom_gray: Grayscale with tunable weights + red/green enhancement
+    """
+    # 1. Read image and validate
+    # img_bgr = cv2.imread(image_path)
+    img_bgr = img
+
+    if img_bgr is None:
+        raise ValueError("Failed to read image - check path!")
+    
+    # 2. Convert BGR (OpenCV default) to RGB and split channels
+    img_rgb = cv2.cvtColor(img_bgr, cv2.COLOR_BGR2RGB)
+    # Convert to float32 to avoid integer overflow/precision loss
+    r = img_rgb[:, :, 0].astype(np.float32)
+    g = img_rgb[:, :, 1].astype(np.float32)
+    b = img_rgb[:, :, 2].astype(np.float32)
+
+    # 3. Calculate DEFAULT grayscale (standard weights)
+    default_gray = 0.299 * r + 0.587 * g + 0.114 * b
+
+    # 4. Calculate CUSTOM grayscale (tunable weights)
+    custom_gray_base = r_weight * r + g_weight * g + b_weight * b
+
+    # 5. Enhance red/green contrast (optional but effective)
+    # Define red/green regions using RGB value ratios (no HSV needed)
+    # red_mask = (r > g * 1.2) & (r > b * 1.2) & (r >= 40)  # Red-dominant pixels
+    # green_mask = (g > r * 1.2) & (g > b * 1.2) & (g >= 40)  # Green-dominant pixels
+    
+    # Apply targeted adjustments to red/green regions
+    custom_gray = custom_gray_base.copy()
+    # custom_gray[red_mask] *= red_boost       # Boost red region grayscale
+    # custom_gray[green_mask] *= green_reduce  # Reduce green region grayscale
+
+    # 6. Normalize to 0-255 range and convert back to uint8 (OpenCV standard)
+    # Clip values first to avoid overflow, then scale if needed
+    default_gray = np.clip(default_gray, 0, 255).astype(np.uint8)
+    custom_gray = np.clip(custom_gray, 0, 255).astype(np.uint8)
+
+    return default_gray, custom_gray
+
+# -------------------------- Test the Code --------------------------
+if __name__ == "__main__":
+    # Replace with your image path
+    IMAGE_PATH = "./data/new_texture.jpg"
+    img = cv2.imread(IMAGE_PATH)
+    dx = -80
+    dy = 0
+    tile = img[410+dy:650+dy,940+dx:1180+dx,:]
+    
+    # Example 1: Use standard weights (verify correctness)
+    default_gray, red_gray = custom_gray_conversion(
+        tile, 
+        r_weight=1, g_weight=0, b_weight=0
+    )
+
+    default_gray, green_gray = custom_gray_conversion(
+        tile, 
+        r_weight=0, g_weight=1, b_weight=0
+    )
+    default_gray, blue_gray = custom_gray_conversion(
+        tile, 
+        r_weight=0, g_weight=0, b_weight=1
+    )
+
+    edge_strength, edge_density = calculate_edge_strength_density(default_gray)
+    local_contrast = calculate_local_contrast(default_gray)
+    print(f"标准灰度: 边缘强度: {edge_strength}, 边缘密度: {edge_density}, 局部对比度: {local_contrast}")
+    edge_strength, edge_density = calculate_edge_strength_density(red_gray)
+    local_contrast = calculate_local_contrast(red_gray)
+    print(f"红色灰度: 边缘强度: {edge_strength}, 边缘密度: {edge_density}, 局部对比度: {local_contrast}")
+    edge_strength, edge_density = calculate_edge_strength_density(green_gray)
+    local_contrast = calculate_local_contrast(green_gray)
+    print(f"绿色灰度: 边缘强度: {edge_strength}, 边缘密度: {edge_density}, 局部对比度: {local_contrast}")
+    edge_strength, edge_density = calculate_edge_strength_density(blue_gray)
+    local_contrast = calculate_local_contrast(blue_gray)
+    print(f"蓝色灰度: 边缘强度: {edge_strength}, 边缘密度: {edge_density}, 局部对比度: {local_contrast}")
+
+    plt.subplot(2, 2, 1)
+    plt.imshow(default_gray, cmap='gray')
+    plt.subplot(2, 2, 2)
+    plt.imshow(red_gray, cmap='gray')
+    plt.subplot(2, 2, 3)
+    plt.imshow(green_gray, cmap='gray')
+    plt.subplot(2, 2, 4)
+    plt.imshow(blue_gray, cmap='gray')
+    plt.show()

src/orisys/Stitcher.py

@@ -446,7 +446,7 @@ class Stitcher:
         n_boxes = 0
         bk_thresh = 0  # initial black threshold
         bk_thresh_step = 5
-        bk_thresh_max = 100
+        bk_thresh_max = 125
 
         # Also include a low global V threshold to capture colored but dark tiles
         hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
@@ -503,6 +503,23 @@ class Stitcher:
             cropped_boxes.append((nx0, ny0, nx1, ny1))
 
         # sort boxes left-to-right, top-to-bottom (useful for grid)
+        # Use quantized y-coordinates to group boxes into rows, making sorting robust to noise
+        if len(cropped_boxes) > 0:
+            # Calculate row tolerance based on typical box height (use median height)
+            heights = [b[3] - b[1] for b in cropped_boxes]
+            median_height = np.median(heights) if heights else target_h
+            # Use a fraction of the box height as tolerance (e.g., 20% of box height)
+            row_tolerance = max(1, int(median_height * 0.2))
+            
+            # Quantize y-coordinates to group boxes into rows
+            def get_row_key(box):
+                y = box[1]  # y0 coordinate
+                # Round y to nearest multiple of row_tolerance to group into rows
+                quantized_y = (y // row_tolerance) * row_tolerance
+                return (quantized_y, box[0])  # Sort by quantized row, then by x
+            
+            cropped_boxes = sorted(cropped_boxes, key=get_row_key)
+        else:
             cropped_boxes = sorted(cropped_boxes, key=lambda b: (b[1], b[0]))
         
         self.boxes = cropped_boxes