2D彩色条码解码器开发：色块间距自动识别难题求助

Great job getting the core encoding and decoding workflow up and running! The fixed-step sampling issue makes total sense—let's tweak the decoder to automatically detect block boundaries so it works reliably no matter how many characters (and color blocks) you're working with.

Why Fixed Step Sizes Fail #

Your current code samples every 200 pixels starting at 150, which works perfectly for exactly 5 blocks (since 1024/5 ≈ 204). But when the number of characters changes, block widths shift, and that fixed step will miss the center of each block. Instead, we need to:

1. Identify exactly where one color block ends and the next begins (boundaries)
2. Sample the center of each detected block to get its corresponding letter

Modified Decode Callback Code #

Here's the updated decode_Callback function with automatic boundary detection:

function decode_Callback(hObject, eventdata, handles)
global img

% First, verify an image was uploaded
if ~exist('img', 'var') || isempty(img)
    errordlg('Please upload an image before decoding!', 'Missing Image');
    return;
end

% Stick with the 500th row for sampling (matches your original approach)
sample_row = 500;
row_rgb = img(sample_row, :, :);
num_pixels = size(row_rgb, 2);

% Step 1: Label each pixel with its color code (1=A, 2=B, 3=C, 4=D)
color_labels = NaN(1, num_pixels);
for col = 1:num_pixels
    r = row_rgb(1, col, 1);
    g = row_rgb(1, col, 2);
    b = row_rgb(1, col, 3);
    
    if r == 255 && g == 0 && b == 0
        color_labels(col) = 1; % Red = A
    elseif r == 0 && g == 255 && b == 0
        color_labels(col) = 2; % Green = B
    elseif r == 255 && g == 255 && b == 0
        color_labels(col) = 3; % Yellow = C
    elseif r == 0 && g == 0 && b == 255
        color_labels(col) = 4; % Blue = D
    end
end

% Step 2: Find boundaries where the color changes
label_changes = find(diff(color_labels) ~= 0);
% Add start (0) and end (full image width) to complete the boundary list
block_boundaries = [0, label_changes, num_pixels];

% Step 3: Sample the center of each block and build the decoded text
letters = 'ABCD';
decoded_text = '';

for i = 1:length(block_boundaries)-1
    % Calculate the middle column of the current block
    block_center = floor((block_boundaries(i) + block_boundaries(i+1)) / 2);
    % Grab RGB values at the block's center
    r_center = img(sample_row, block_center, 1);
    g_center = img(sample_row, block_center, 2);
    b_center = img(sample_row, block_center, 3);
    
    % Map center RGB to corresponding letter
    if r_center == 255 && g_center == 0 && b_center == 0
        decoded_text = [decoded_text, 'A'];
    elseif r_center == 0 && g_center == 255 && b_center == 0
        decoded_text = [decoded_text, 'B'];
    elseif r_center == 255 && g_center == 255 && b_center == 0
        decoded_text = [decoded_text, 'C'];
    elseif r_center == 0 && g_center == 0 && b_center == 255
        decoded_text = [decoded_text, 'D'];
    end
end

% Update the output text box with the result
set(handles.out, 'string', decoded_text);
end

How This Works #

• Color Labeling: We first tag every pixel in the sample row with a code that maps to its corresponding letter (A-D).
• Boundary Detection: By looking for where the color label changes, we pinpoint exact block edges. We add the start and end of the image to make sure we cover all blocks.
• Center Sampling: For each block, we calculate its middle point (the most reliable spot to sample, avoiding edge artifacts) and map that RGB value back to the correct letter.

Optional Improvements #

• Add error handling for unexpected colors (e.g., show a warning if a pixel doesn't match A-D's RGB values)
• Use flexible color matching instead of exact == checks (e.g., abs(r-255) < 10) to account for minor image compression artifacts

内容的提问来源于stack exchange，提问作者Nitin Tiwari