View Categories

align

Overview #

This example explains how to align depth and color streams captured from different viewpoints, and highlights potential issues such as sampling artifacts and occlusion.

Expect Output #

Picture1 | LIPS® Developer Documentation

Prerequisites #

Tutorial #

C++
Mat colorAligned, depthAligned;
if (align_mode == 0) {
          
   if (depthColor.size() != colorBGR.size()) {
                    resize(depthColor, depthAligned, colorBGR.size(), 0, 0, INTER_NEAREST);
                }
                else {
                    depthAligned = depthColor;
                }
                colorAligned = colorBGR.clone();
            }
            else {
                if (colorBGR.size() != depthColor.size()) {
                    resize(colorBGR, colorAligned, depthColor.size(), 0, 0, INTER_NEAREST);
                }
                else {
                    colorAligned = colorBGR;
                }
                depthAligned = depthColor.clone();
            }
  • Implements two alignment strategies controlled by align_mode:
    • Mode 0 (Depth→Color): resize the depth image to match the color resolution.
    • Mode 1 (Color→Depth): resize the color image to match the depth resolution.
  • Uses INTER_NEAREST for resizing to avoid creating non-existent values in depth-derived data.
  • If hardware registration (regEnabled) is ON and resolutions already match, the resize becomes a no-op (identity mapping).
C++
double alpha = alpha_slider / 100.0;      // color weight
            double beta = 1.0 - alpha;               // depth weight

            Mat blended;
            addWeighted(colorAligned, alpha, depthAligned, beta, 0.0, blended);

            // ----------- Add text overlay -----------
            string modeText = (align_mode == 0) ?
                "Mode: Depth -> Color (output size = Color)" :
                "Mode: Color -> Depth (output size = Depth)";
            putText(blended, modeText, Point(10, 25),
                FONT_HERSHEY_SIMPLEX, 0.7, Scalar(0, 255, 0), 2);

            char alphaTxt[64];
            snprintf(alphaTxt, sizeof(alphaTxt), "Color Alpha: %.0f %%", alpha * 100.0);
            putText(blended, alphaTxt, Point(10, 55),
                FONT_HERSHEY_SIMPLEX, 0.7, Scalar(0, 255, 255), 2);

            if (!regEnabled) {
                putText(blended, "Warning: Registration OFF (simple resize only)",
                    Point(10, 85),
                    FONT_HERSHEY_SIMPLEX, 0.6, Scalar(0, 0, 255), 2);
            }
  • Uses addWeighted to overlay color and depth:
    • alpha = contribution of the color image.
    • beta = contribution of the depth image.
  • Displays the current alignment mode and alpha value as on-screen text.

Updated on 2026 年 4 月 1 日