Showcase (Web Version)

How to run

For the cli version :

cargo run -p image_cli

For the web version :

cargo run -p image_web

Things to add

1. Spatial/Convolutional Features

• apply_gaussian_blur(kernel_size: usize, sigma: f32): Necessary for noise reduction and smoothing.
• apply_sobel_operator(): Extracts edge magnitude and direction features (crucial for object detection).
• apply_laplacian_filter(): Used for blob detection and image sharpening.
• apply_box_filter(kernel_size: usize): A faster alternative to Gaussian blur for general feature smoothing.

2. Statistical & Morphological Features

• get_image_entropy(): Measures the complexity/information density of the image.
• get_image_variance(): A core metric for contrast analysis (feature distinctiveness).
• get_image_skewness_kurtosis(): Provides data on the distribution shape of pixel intensities.
• apply_threshold(value: f32): Generates binary feature masks (segmentation).
• apply_dilation_erosion(): Morphological operations to clean up masks or extract shapes.

3. Normalization & Preprocessing (For ML Pipelines)

• normalize_z_score(): Shifts the data to have a mean of 0 and std-dev of 1 (essential for Neural Network input).
• rescale_image(new_width: u32, new_height: u32): Standardizes image dimensions for model compatibility (interpolation).
• clip_outliers(min: f32, max: f32): Prevents extreme pixel values from affecting ML training convergence.
• flatten(): Converts the RawImage data into a 1D vector (or feature array) ready for model input.

4. Image Quality/Metric Features

• get_contrast_ratio(): Specifically, the Michelson or RMS contrast.
• get_luminance_map(): Extracts the Y channel (from YCbCr conversion) as a feature, which is often more useful for models than raw RGB.
• compute_mse(img1: &RawImage, img2: &RawImage): Mean Squared Error, used to calculate reconstruction loss in Autoencoders or compression quality.

5. Advanced Geometry/Structure

• get_bounding_box(threshold: f32): Finds the minimal rectangle enclosing non-zero features.
• calculate_centroid(): Returns the center of mass of the image intensities (useful for object localization).
• apply_affine_transform(matrix: [[f32; 3]; 2]): For image augmentation (scaling, shearing, translation).

6. Safety & Debugging Additions

• is_valid_coord(x: u32, y: u32): A helper to resolve the TODO regarding out-of-bounds errors before attempting to access pixel data.
• to_grayscale_luminance_weighted(): Instead of simple averaging, use the perceptually accurate formula: 0.299R + 0.587G + 0.114B.