CV Feature Enhancement Processor

A command-line tool for batch processing images with various computer vision feature enhancement techniques. This tool is designed to help calibrate and test different CV preprocessing methods by applying them to a set of source images.

Features

• Process all images in a source directory with various CV-specific transformations
• Multiple edge detection methods (Canny, Sobel, Laplacian, structured edges)
• Feature enhancement techniques (CLAHE, adaptive thresholding)
• Morphological operations for feature manipulation
• Corner and blob detection enhancement
• Parallel processing support for faster execution
• Progress bar to track processing status
• Organizes processed images in timestamped subfolders for easy tracking and comparison

Sample Results

Here's a demonstration of various transformations applied to a sample image:

Original Image

Original input image

Edge Detection Examples

Canny edge detection with thresholds (50, 150) after Gaussian blur (5x5)

Advanced edge detection combining Sobel gradients with non-maximum suppression and double thresholding

Feature Enhancement

CLAHE (Contrast Limited Adaptive Histogram Equalization) applied in LAB color space, clipLimit=2.0, tileGrid=8x8

Adaptive Gaussian thresholding with block size=11 and C=2

Feature Detection

Harris corner detection (blockSize=2, ksize=3, k=0.04) with corners highlighted in red

Blob detection using SimpleBlobDetector with minThreshold=10, maxThreshold=200, minArea=100

Installation

1. Clone this repository
2. Create and activate a conda environment (recommended):

conda create -n cv_imgprocessor python=3.8
conda activate cv_imgprocessor

3. Install dependencies:

pip install -r requirements.txt

Usage

Basic Usage

Place your source images in the img_source directory and run:

python img_processor.py

This will apply all available transformations to each image and save the results in a timestamped subfolder within the img_output directory (e.g., img_output/processed_20230619_142530/).

Command Line Options

python img_processor.py [-h] [-s SOURCE] [-o OUTPUT] [-t TRANSFORM [TRANSFORM ...]]
                        [-c COMBO [COMBO ...]] [-w WORKERS] [-l]

Arguments:

• -h, --help: Show help message
• -s, --source: Source directory (default: img_source)
• -o, --output: Output directory (default: img_output)
• -t, --transform: Apply specific transformations
• -c, --combo: Apply specific combination transformations
• -w, --workers: Number of worker processes (default: 1)
• -l, --list: List available transformations and exit

Examples

Apply all transformations to all images:

python img_processor.py

Apply specific edge detection methods:

python img_processor.py --transform canny_edges structured_edges

Apply feature extraction combination:

python img_processor.py --combo feature_extraction

List available transformations:

python img_processor.py --list

Process with multiple worker processes:

python img_processor.py --workers 4

Available Transformations

Edge Detection Methods

• canny_edges: Standard Canny edge detection
• canny_edges_tight: Canny with tighter thresholds for stronger edges
• canny_edges_loose: Canny with looser thresholds for weaker edges
• sobel_edges: Sobel gradient-based edge detection
• laplacian_edges: Laplacian edge detection
• structured_edges: Advanced edge detection combining multiple methods
• ridge_detection: Ridge detection using Hessian matrix eigenvalues

Feature Enhancement

• clahe_enhance: Contrast Limited Adaptive Histogram Equalization
• adaptive_threshold: Adaptive Gaussian thresholding
• otsu_threshold: Otsu's automatic thresholding

Morphological Operations

• dilate_features: Dilation to expand features
• erode_features: Erosion to shrink features
• open_features: Opening operation (erosion followed by dilation)
• close_features: Closing operation (dilation followed by erosion)
• tophat_features: Top-hat transformation for bright features on dark background

Feature Detection

• corner_enhance: Harris corner detection enhancement
• blob_enhance: Blob detection and enhancement

Combination Transformations

• feature_extraction: Combines CLAHE and structured edges for optimal feature extraction
• blob_detection: Enhances and detects blob-like structures
• corner_detection: Enhances and detects corner features
• edge_analysis: Multi-step edge detection and enhancement
• texture_analysis: Enhances texture features using multiple methods