ImageFilterApp.py

import streamlit as st
import cv2
import numpy as np
from PIL import Image

st.set_page_config(page_title="🖼️ Image Blurring & Sharpening App", layout="centered")

st.title("🧪 Image Filtering Playground")
st.markdown("""
Welcome to the **Image Blurring and Sharpening App**!  
Here you can upload an image and explore how different filters affect it.  
🔍 _Great for learning computer vision basics!_
""")

uploaded_file = st.file_uploader("📤 Upload an image", type=["jpg", "png", "jpeg"])

if uploaded_file:
    file_bytes = np.asarray(bytearray(uploaded_file.read()), dtype=np.uint8)
    img = cv2.imdecode(file_bytes, 1)

    st.markdown("## 🛠️ Choose a Filter")
    filter_type = st.selectbox(
        "Select Filter", ["None", "Gaussian Blur", "Median Blur", "Sharpening"]
    )

    result_img = img.copy()

    if filter_type == "Gaussian Blur":
        st.info(
            "🌀 Gaussian Blur: Averages neighboring pixel values to smooth image and remove noise."
        )
        kx = st.slider("Kernel Size X", 1, 31, 3, step=2)
        ky = st.slider("Kernel Size Y", 1, 31, 3, step=2)
        sigma = st.slider("Sigma (Blur Strength)", 0.0, 5.0, 1.0)
        result_img = cv2.GaussianBlur(img, (kx, ky), sigma)

    elif filter_type == "Median Blur":
        st.info(
            "🎯 Median Blur: Uses the **median** of surrounding pixels — great for salt & pepper noise!"
        )
        k = st.slider("Kernel Size", 1, 31, 3, step=2)
        result_img = cv2.medianBlur(img, k)

    elif filter_type == "Sharpening":
        st.info("✏️ Sharpening Filter: Enhances edges and brings out details.")
        ddepth_map = {
            "Same as input (-1)": -1,
            "8-bit unsigned (CV_8U)": cv2.CV_8U,
            "16-bit signed (CV_16S)": cv2.CV_16S,
            "32-bit float (CV_32F)": cv2.CV_32F,
        }
        ddepth_label = st.selectbox("Select Output Depth", list(ddepth_map.keys()))
        ddepth = ddepth_map[ddepth_label]
        kernel = np.array([[0, -1, 0], [-1, 5, -1], [0, -1, 0]])

        result_img = cv2.filter2D(img, ddepth, kernel)

        # Handle visualization of higher-depth outputs (CV_16S, CV_32F)
        if ddepth in [cv2.CV_16S, cv2.CV_32F]:
            # Normalize to 0–255 and convert back to uint8
            norm_img = cv2.normalize(result_img, None, 0, 255, cv2.NORM_MINMAX)
            result_img = np.uint8(norm_img)

    if filter_type != "None":
        st.markdown("### 🔍 Original vs. Processed Image")
        col1, col2 = st.columns(2)

        with col1:
            st.image(img[:, :, ::-1], caption="Original", use_container_width=True)

        with col2:
            st.image(
                result_img[:, :, ::-1],
                caption=f"{filter_type} Applied",
                use_container_width=True,
            )

        result_pil = Image.fromarray(cv2.cvtColor(result_img, cv2.COLOR_BGR2RGB))
        st.download_button(
            "📥 Download Processed Image",
            data=result_pil.tobytes(),
            file_name="processed_image.jpg",
            mime="image/jpeg",
        )
    else:
        st.image(img[:, :, ::-1], caption="Original Image", use_container_width=True)

    st.markdown("---")
    st.caption("Built with ❤️ using OpenCV + Streamlit")

else:
    st.warning("👆 Upload an image to begin.")