CVForge

A lightweight Computer Vision CLI toolkit written in Go

A modular image processing framework featuring classic computer vision algorithms implemented from scratch in pure Go.

---

Overview

CVForge is a lightweight Computer Vision command-line toolkit implemented entirely in Go.

Unlike wrappers around OpenCV, CVForge focuses on implementing classical image processing algorithms from scratch while maintaining a clean, modular architecture.

It is designed for both practical image processing and educational purposes.

---

Features

Geometry

• Resize
• Rotate
• Crop

Color Processing

• Grayscale
• Histogram Equalization

Blur

• Box Blur
• Gaussian Blur
• Median Blur

Image Enhancement

• Sharpen
• Emboss

Edge Detection

• Sobel Edge Detection

Thresholding

• Binary
• Binary Inverse
• Truncate
• To Zero
• To Zero Inverse

Morphology

• Erosion
• Dilation
• Opening
• Closing

---

Example Results

Original

Original

---

Basic Filters

Grayscale	Gaussian	Median

---

Image Enhancement

Sharpen	Emboss	Equalize

---

Edge Detection

Sobel

---

Threshold

Binary	Binary Inv

Truncate	To Zero	To Zero Inv

---

Morphology

Structuring Element Comparison

Operation	Square	Cross
Erosion
Dilation
Opening
Closing

---

Architecture

CVForge adopts a modular filter pipeline architecture.

Each image processing operation is implemented as an independent filter.

             +-------------+
Input Image ->| Load Image |
             +-------------+
                    │
                    ▼
           +-----------------+
           | Filter Pipeline |
           +-----------------+
              │      │
              │      ├── Grayscale
              │      ├── Gaussian
              │      ├── Sobel
              │      ├── Threshold
              │      └── ...
                    ▼
             +-------------+
             | Save Image  |
             +-------------+

Every filter implements the same interface:

type Filter interface {
    Process(
        image.Image,
        progress.Reporter,
    ) image.Image
}

---

Project Structure

CVForge
├── assets/
├── cmd/
│   ├── blur.go
│   ├── crop.go
│   ├── dilation.go
│   ├── emboss.go
│   ├── equalize.go
│   ├── erosion.go
│   ├── gaussian.go
│   ├── grayscale.go
│   ├── median.go
│   ├── opening.go
│   ├── closing.go
│   ├── resize.go
│   ├── rotate.go
│   ├── sobel.go
│   └── threshold.go
│
├── internal/
│   ├── filters/
│   ├── histogram/
│   ├── imageio/
│   ├── imageutil/
│   ├── kernel/
│   ├── morphology/
│   ├── pipeline/
│   ├── progress/
│   └── service/
│
├── examples/
└── main.go

---

Supported Commands

Command	Description
grayscale	Convert image to grayscale
blur	Box blur
gaussian	Gaussian blur
median	Median blur
resize	Resize image
rotate	Rotate image
crop	Crop image
sharpen	Sharpen image
emboss	Emboss effect
sobel	Sobel edge detection
threshold	Binary thresholding
equalize	Histogram equalization
erosion	Morphological erosion
dilation	Morphological dilation
opening	Morphological opening
closing	Morphological closing

---

Installation

Clone the repository

git clone https://github.com/kylin419/CVforge.git
cd CVforge

Build

go build

---

Usage

Grayscale

cvforge grayscale input.jpg output.jpg

Gaussian Blur

cvforge gaussian input.jpg output.jpg \
    --size 5 \
    --sigma 1.5

Median Blur

cvforge median input.jpg output.jpg \
    --radius 2

Resize

cvforge resize input.jpg output.jpg \
    --w 800 \
    --h 600

Rotate

cvforge rotate input.jpg output.jpg \
    --angle 90

Crop

cvforge crop input.jpg output.jpg \
    --x 100 \
    --y 50 \
    --w 400 \
    --h 300

Threshold

cvforge threshold input.jpg output.jpg \
    --value 127 \
    --type binary

Available threshold types:

• binary
• binary-inv
• truncate
• tozero
• tozero-inv

Histogram Equalization

cvforge equalize input.jpg output.jpg

Erosion

cvforge erosion input.png output.png \
    --element square

Dilation

cvforge dilation input.png output.png \
    --element square

Opening

cvforge opening input.png output.png \
    --element square

Closing

cvforge closing input.png output.png \
    --element square

Supported structuring elements:

• square
• cross

---

Algorithms

Filtering

• Box Blur
• Gaussian Blur
• Median Blur

Convolution

• Generic Convolution Engine
• Dynamic Gaussian Kernel
• Preset Kernels

Enhancement

• Sharpen
• Emboss
• Histogram Equalization

Edge Detection

• Sobel Operator

Thresholding

• Binary
• Binary Inverse
• Truncate
• To Zero
• To Zero Inverse

Morphology

• Binary Erosion
• Binary Dilation
• Opening
• Closing
• Configurable Structuring Elements

---

Design Goals

• Pure Go implementation
• No OpenCV dependency
• Modular filter pipeline
• Reusable convolution engine
• Reusable morphology engine
• Easy to extend
• Educational implementation of classical computer vision algorithms

---

Performance

• Pure Go implementation
• Streaming image processing pipeline
• Progress reporting for long-running operations
• No external computer vision libraries

---

Roadmap

v1.3 ✅

• Geometry Transformations
• Grayscale
• Generic Convolution Engine
• Gaussian Blur
• Median Blur
• Sharpen
• Emboss
• Sobel Edge Detection
• Histogram Equalization
• Threshold Family

v1.4 ✅

• Morphological Operations
  • Erosion
  • Dilation
  • Opening
  • Closing

v1.5

• Canny Edge Detection
• Histogram Visualization
• CLAHE

v2.0

• Harris Corner Detection
• Hough Transform
• Template Matching

---

License

MIT License

---

Author

Kylin

GitHub: https://github.com/kylin419