TN12-A deep learning into confidence filter

Written By:Sophia.feng Updated: 2024-10-2

What is confidence filter?

Confidence filter a high-pass filter. Confidence data can be used for filtering because the value of confidence indicates the trustworthiness of the signal-to-noise ratio of energy reception. The higher confidence value refers the data is more trustworthy. So we can set a confidence threshold to filter out the low confidence pixels by marking it as invalid. And the higher the threshold is, the more pixels will be marked as invalid.
For Time-of-Flight(ToF) camera, the confidence value mainly relates to the reflectance of the object, in other word, it can be used to filter out low reflectance objects.

Example of confidence filter application on Vzense ScepterGUITool

Users can try the confidence filter function on Scepter GUITool, please see below figure:

Detect the object with time of flight camera, turn off the filter, and its point cloud map are shown below:

The confidence filter threshold ranges are [1, 100], turn on the confidence filter. The larger the threshold set by the user, the points with worse signal quality will be removed.
The below point cloud shows the samples with confidence filter threshold value is 6:

The below point cloud shows the samples with confidence filter threshold value is 42:

Pros and cons of confidence filter

Pros: The points with poor signal quality are eliminated.
Cons: The larger threshold value of confidence filtering may cause data loss on the depth map.

Related Posts

Welcome to visit!

USA Office Address: 5451 Great America Parkway, Suite 301,Santa Clara, CA 95054
( Please refer to the map on the right->)

Headquaters Address: Building F, Innovation Park II, No. 1, Keyuan Wei 1st Road, Laoshan District, Qingdao, Shandong, China

Contact Now