remove-background

Overview

This example shows you how to remove the background of an image using depth camera with range filter.

Expect Output

Prerequisite

• OpenCV Viewer
• Align Depth & Color
• Range Filter

Tutorial

Modify from range-filter, we add two variables to determinate the desired distance. Then add two keyboard shorcut to set these values.

thresholdValueUp = 530
thresholdValueDown = 385

input = cv2.waitKey(1)
if input == ord('p'):
    if thresholdValueUp + 5 <= 2000:
        thresholdValueUp += 5
    else:
        thresholdValueUp = 2000
    print(f"Threshold Value (Up): {thresholdValueUp} (Down): {thresholdValueDown}")

elif input == ord('o'):
    if thresholdValueUp - 5 >= 0 and thresholdValueUp - 5 >= thresholdValueDown:
        thresholdValueUp -= 5
    print(f"Threshold Value (Up): {thresholdValueUp} (Down): {thresholdValueDown}")

elif input == ord('x'):
    if thresholdValueDown + 5 <= 2000 and thresholdValueDown + 5 <= thresholdValueUp:
        thresholdValueDown += 5
    print(f"Threshold Value (Up): {thresholdValueUp} (Down): {thresholdValueDown}")

elif input == ord('z'):
    if thresholdValueDown - 5 >= 0:
        thresholdValueDown -= 5
    else:
        thresholdValueDown = 0
    print(f"Threshold Value (Up): {thresholdValueUp} (Down): {thresholdValueDown}")

elif input == ord('q'):
    break

Using these two values to create range filter mask image like we did before.

 _, thres = cv2.threshold(thres, thresholdValueDown, 1024, cv2.THRESH_TOZERO)
_, thres = cv2.threshold(thres, thresholdValueUp, 1024, cv2.THRESH_TOZERO_INV)
_, thres = cv2.threshold(thres, 1, 1024, cv2.THRESH_BINARY)

Then we use OpenCV bitwise_and operation on the color image and the mask. We also need to convert the mask to CV_8UC1 because that the format needed by bitwise_and. Finally we will get the result image that remove the background.

thres = cv2.convertScaleAbs(thres, alpha=255.0 / 1024.0)
res = cv2.bitwise_and(rgbMat, rgbMat, mask=thres)

Full code

remove-background.py