app.py

import cv2
import numpy as np
import json
import gradio as gr
import matplotlib.pyplot as plt

# === Helper: Rotated rectangle corners ===
def get_rotated_rect_corners(x, y, w, h, rotation_deg):
    rot_rad = np.deg2rad(rotation_deg)
    cos_r = np.cos(rot_rad)
    sin_r = np.sin(rot_rad)

    R = np.array([[cos_r, -sin_r],
                  [sin_r,  cos_r]])

    cx = x + w/2
    cy = y + h/2

    local_corners = np.array([
        [-w/2, -h/2],
        [ w/2, -h/2],
        [ w/2,  h/2],
        [-w/2,  h/2]
    ])

    rotated_corners = np.dot(local_corners, R.T)
    corners = rotated_corners + np.array([cx, cy])
    return corners.astype(np.float32)

# === Preprocessing: Grayscale + CLAHE ===
def preprocess_gray_clahe(img):
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    clahe = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8, 8))
    return clahe.apply(gray)

# === Detect and match features ===
def detect_and_match(img1_gray, img2_gray, method="SIFT", ratio_thresh=0.78):
    if method == "SIFT":
        sift = cv2.SIFT_create(nfeatures=5000)
        kp1, des1 = sift.detectAndCompute(img1_gray, None)
        kp2, des2 = sift.detectAndCompute(img2_gray, None)
        matcher = cv2.BFMatcher(cv2.NORM_L2)
    elif method == "ORB":
        orb = cv2.ORB_create(5000)
        kp1, des1 = orb.detectAndCompute(img1_gray, None)
        kp2, des2 = orb.detectAndCompute(img2_gray, None)
        matcher = cv2.BFMatcher(cv2.NORM_HAMMING)
    elif method == "BRISK":
        brisk = cv2.BRISK_create()
        kp1, des1 = brisk.detectAndCompute(img1_gray, None)
        kp2, des2 = brisk.detectAndCompute(img2_gray, None)
        matcher = cv2.BFMatcher(cv2.NORM_HAMMING)
    elif method == "KAZE":
        kaze = cv2.KAZE_create()
        kp1, des1 = kaze.detectAndCompute(img1_gray, None)
        kp2, des2 = kaze.detectAndCompute(img2_gray, None)
        matcher = cv2.BFMatcher(cv2.NORM_L2)
    elif method == "AKAZE":
        akaze = cv2.AKAZE_create()
        kp1, des1 = akaze.detectAndCompute(img1_gray, None)
        kp2, des2 = akaze.detectAndCompute(img2_gray, None)
        matcher = cv2.BFMatcher(cv2.NORM_HAMMING)
    else:
        return None, None, None

    raw_matches = matcher.knnMatch(des1, des2, k=2)
    good = []
    for m, n in raw_matches:
        if m.distance < ratio_thresh * n.distance:
            good.append(m)
    return kp1, kp2, good

# === Main function for Gradio ===
def homography_demo(flat_file, persp_file, json_file):
    flat_img = cv2.imdecode(np.frombuffer(flat_file.read(), np.uint8), cv2.IMREAD_COLOR)
    persp_img = cv2.imdecode(np.frombuffer(persp_file.read(), np.uint8), cv2.IMREAD_COLOR)
    mockup = json.load(json_file)

    roi_data = mockup["printAreas"][0]["position"]
    roi_x = roi_data["x"]
    roi_y = roi_data["y"]
    roi_w = mockup["printAreas"][0]["width"]
    roi_h = mockup["printAreas"][0]["height"]
    roi_rot_deg = mockup["printAreas"][0]["rotation"]

    flat_gray = preprocess_gray_clahe(flat_img)
    persp_gray = preprocess_gray_clahe(persp_img)

    methods = ["SIFT", "ORB", "BRISK", "KAZE", "AKAZE"]
    outputs = []

    for method in methods:
        kp1, kp2, good_matches = detect_and_match(flat_gray, persp_gray, method=method)

        src_pts = np.float32([kp1[m.queryIdx].pt for m in good_matches]).reshape(-1, 1, 2)
        dst_pts = np.float32([kp2[m.trainIdx].pt for m in good_matches]).reshape(-1, 1, 2)

        H, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0)

        roi_corners_flat = get_rotated_rect_corners(roi_x, roi_y, roi_w, roi_h, roi_rot_deg)
        roi_corners_persp = cv2.perspectiveTransform(roi_corners_flat.reshape(-1, 1, 2), H).reshape(-1, 2)

        persp_debug = persp_img.copy()
        cv2.polylines(persp_debug, [roi_corners_persp.astype(int)], True, (0, 255, 0), 2)
        for (px, py) in roi_corners_persp:
            cv2.circle(persp_debug, (int(px), int(py)), 5, (255, 0, 0), -1)

        # Convert BGR to RGB for Gradio display
        persp_debug_rgb = cv2.cvtColor(persp_debug, cv2.COLOR_BGR2RGB)
        outputs.append((persp_debug_rgb, method))

    return outputs

# === Gradio Interface ===
with gr.Blocks() as demo:
    gr.Markdown("## Homography ROI Demo with Multiple Feature Detectors")
    with gr.Row():
        flat_input = gr.File(label="Upload Flat Image")
        persp_input = gr.File(label="Upload Perspective Image")
        json_input = gr.File(label="Upload mockup.json")
    output_gallery = gr.Gallery(label="Perspective ROI Results").style(grid=[2], height="400px")
    run_btn = gr.Button("Run Homography")

    run_btn.click(homography_demo, 
                  inputs=[flat_input, persp_input, json_input],
                  outputs=output_gallery)

demo.launch()