haar

Haar

This project provides a Golang implementation of the Haar 2D transform, following an approach similar to imgSeek. The library can generate image hashes for indexing and similarity search in iqdb.

[!WARNING] Output from the Go package shows small deviations compared to the C reference. This does not affect the final result, and all images are successfully found. Refer to tests for specifics.

Installation

go get github.com/ezh0v/haar

Quick Start

Create your .go file. For example: main.go.

package main

import (
	"fmt"
	"log"
	"os"

	"github.com/ezh0v/haar"
)

func main() {
	image, err := os.ReadFile("your_image_file.jpg")
	if err != nil {
		log.Fatal(err)
	}

	signature, err := haar.SignatureFromBlob(image)
	if err != nil {
		log.Fatal(err)
	}

	fmt.Println(signature)
}

Run your program:

go run main.go

Motivation

Most booru sites need an image file to search. Danbooru can use an image's hash instead, which is much faster.