ImageCompresion

A project for compresing images using SVD algorithm. The script splits the image into its red, green, and blue color channels, applies SVD to each channel, and then reconstructs the image with reduced dimensions. This compression technique can significantly reduce the size of the image while preserving its visual quality.

Tabel of content

• Requirements
  
• Code Explanation
  
• Usage
  

Requirements

This project uses Numpy and PIL libraries.

Code Explanation

The compress_color_image function performs the following steps:

1. Load Images:

image = Image.open(image_path)
A = np.array(image)

2. Split color channels

R = A[:,:,0]
G = A[:,:,1]
B = A[:,:,2]

3. Compress each channel using SVG

def compress_channel(channel):
    U, sigma, VT = np.linalg.svd(channel, full_matrices=False)
    U_k = U[:, :k]
    sigma_k = np.diag(sigma[:k])
    VT_k = VT[:k, :]
    return np.dot(U_k, np.dot(sigma_k, VT_k))

R_k = compress_channel(R)
G_k = compress_channel(G)
B_k = compress_channel(B)

4. Reconstruct and Save the image

compressed_image_array = np.stack([R_k, G_k, B_k], axis=2)
compressed_image_array = np.clip(compressed_image_array, 0, 255)
compressed_image = Image.fromarray(compressed_image_array.astype('uint8'))
compressed_image.save(output_path)

Usage

1. Save the code file and move it to the image folder
   
2. Rename the last code line
   
3. Run the code