SBOU157 September   2015 OPT8241

 

  1.   Voxel Viewer
    1.     Trademarks
    2. 1 Introduction
      1. 1.1 TI 3D Depth Sensors
      2. 1.2 Operating Principles
      3. 1.3 Voxel Viewer and Voxel SDK
    3. 2 User Interface
      1. 2.1 Menu Bar
        1. 2.1.1 File
        2. 2.1.2 Settings
          1. 2.1.2.1 Camera Profiles
          2. 2.1.2.2 Edit Profiles
          3. 2.1.2.3 Statistics
        3. 2.1.3 Windows
        4. 2.1.4 Help
      2. 2.2 Source Bar
      3. 2.3 Main Viewport
      4. 2.4 Left Pane
        1. 2.4.1 Side Viewports
      5. 2.5 Right Pane
        1. 2.5.1 Frequently Used Parameters Window
        2. 2.5.2 Parameter List Window
      6. 2.6 Bottom Pane
        1. 2.6.1 Watch List and Statistics Window
        2. 2.6.2 Data Flow Diagram Window
        3. 2.6.3 Logs Window
    4. 3 Basic Operations
      1. 3.1 Connecting Devices
      2. 3.2 Adjusting the Settings
      3. 3.3 Visualizing the Data
        1. 3.3.1 Ambient
        2. 3.3.2 Amplitude
        3. 3.3.3 Depth
        4. 3.3.4 Distance
        5. 3.3.5 Phase
        6. 3.3.6 Point Clouds
        7. 3.3.7 Histogram
      4. 3.4 De-Noising
        1. 3.4.1 Temporal Filters
          1. 3.4.1.1 IIR Filter
          2. 3.4.1.2 Median Filter
        2. 3.4.2 Spatial Filters
          1. 3.4.2.1 Smooth Filter
          2. 3.4.2.2 Bilateral Filter
          3. 3.4.2.3 Median Filters
        3. 3.4.3 Recommended Starting Point
    5. 4 Calibration
      1. 4.1 Lens Calibration
      2. 4.2 Frequency Calibration
      3. 4.3 Crosstalk Calibration
      4. 4.4 Nonlinearity Calibration
      5. 4.5 Temperature Calibration
      6. 4.6 Common Phase Calibration
      7. 4.7 Pixel-Wise Calibration
      8. 4.8 Profiles and Calibration
    6. 5 TFC Programming
    7. 6 Summary
    8. 7 References

3.4.2.3 Median Filters

The spatial median filter, not to be confused with the temporal median filter, is also designed to remove spatial noise; but it is done by ordering the pixels in a square sub-image (kernel) and then taking the median value as the center pixel’s filtered output. Since the median filter uses the measured pixel value rather than interpolated pixel value, the filtered image is generally a better reproduction of the actual scene. The spatial median filter has several properties. The most important of all is the half-kernel size, which defines a kernel of size of 2n + 1. For the example shown in Figure 21, the kernel size is 5. The dead band step and stability should be generally untouched at 0.01 and 0.10, respectively.

fig21_sbou157.gifFigure 21. Spatial Median Filter Properties