Bar code generation and decoding

In this article we will go into detail on how to convert a character string into a black and white barcode (as an image) and vice versa. The code uses the Images package to work with images.

The programme we have implemented consists of two main functions:

1. string_to_barcode - Converts a string into a barcode image;
2. barcode_to_string - decodes the barcode image back to a string.

using Images

String_to_barcode function

Operating logic

1. Initialisation of empty array bits: it stores bits of all characters of the string.

2. Cycle through each character of the string:

   • bitstring(UInt8(char)) converts a character to its 8-bit representation,
   • parse(Int, bit) converts each bit to the corresponding integer,
   • append! adds these numbers to the array bits.

3. Creating a two-dimensional array for an image:

   • bits' transposes a vector of bits (converts to a string),
   • repeat repeats this string length(bits)/2 times to create a two-dimensional array (image) where each row is identical,
   • factor length(bits)/2 is selected to create a rectangular image.

4. Return result: the function returns a two-dimensional array of 0's and 1's, which can be interpreted as a black and white image.

function string_to_barcode(s::String)
    bits = []
    for char in s
        binary_str = bitstring(UInt8(char)) # Получаем бинарное представление символа (8 бит)
        append!(bits, [parse(Int, bit) for bit in binary_str]) # Преобразуем каждый бит в число и добавляем в вектор
    end
    repeated_bits = repeat(bits', Int.(length(bits)/2),1) # Повторяем вектор бит
    return repeated_bits
end

string_to_barcode (generic function with 1 method)

Example of use:

barcode = string_to_barcode("_Engee_") # Переводим строку в битовое представление
img = Gray.(barcode) # Преобразуем в изображение
display(img) # Показываем изображение
save("barcode.png", img) # Сохраняем в файл

Function barcode_to_string

Operating logic

1. Binarising the image:

   • Gray.(img) converts the image to grayscale;
   • .> 0.5 converts to pure black and white (values greater than 0.5 become 1, the rest become 0).

2. Extracting a bit sequence:

   • binary_img[10, :] takes the 10th line of the image (you can take any line, since all lines are the same).

3. Character decoding:

   • loops through the bits in increments of 8 (each character is encoded with 8 bits),
   • for every 8 bits
     • reduce((x, y) -> x * 2 + y, byte) converts an array of bits into a number,
     • Char() converts the number to the corresponding character.

4. String assembly:

   • join(chars) combines all characters into a single string.

function barcode_to_string(img)
    binary_img = Gray.(img) .> 0.5 # Преобразуем изображение в бинарный формат (0 или 1)
    bits = binary_img[10, :] # Получаем первую строку изображения (так как все строки одинаковы)
    # Преобразуем биты в символы
    chars = []
    for i in 1:8:length(bits)
        # Берем 8 бит и преобразуем их в символ
        byte = bits[i:i+7]
        char = Char(reduce((x, y) -> x * 2 + y, byte))
        push!(chars, char)
    end
    # Собираем символы в строку
    return join(chars)
end

barcode_to_string (generic function with 1 method)

Example usage:

img = load("barcode.png") # Загружаем изображение штрихкода
display(img) # Показываем изображение
decoded_string = barcode_to_string(img) # Декодируем строку
println("Decoded string: ", decoded_string) # Выводим результат

Decoded string: _Engee_

Conclusion

This code demonstrates the basic principle of converting a string to a visual representation and back again. This principle can be extended for more complex usage scenarios.

Some implementation features are also worth mentioning.

1. Character encoding: each character is encoded with exactly 8 bits (ASCII), which limits the character set but simplifies the implementation.

2. Visible representation: The barcode is created by repeating the same bit sequence in multiple lines, which makes it more visible.

3. Stability: a specific image line (10th line) is taken during decoding, which makes the algorithm vulnerable to image corruption. Real systems usually use more complex reconstruction logic.