detection_pseudocode.txt

Algorithm 1: Detection
def DETECTION(frame, bk_sub_KNN)
    if frame is None:
        break
    foreground_mask = bk_sub_KNN.apply(frame)
    kernel = np.ones((10,10),np.uint8)
    Apply Opening Morphology to foreground_mask remove very small movements
    Apply Dilation Morphology to increase detection area
    Find All Contours in foreground_mask
    temp_bb_TL_x = max_x
    temp_bb_TL_y = max_y
    temp_bb_BR_x = min_x
    temp_bb_BR_y = min_y
    for contour in Contours:
        x,y,w,h = boundingBox(contour)
        if x < temp_bb_TL_x:            ------------------> Find Region of Interest
            temp_bb_TL_x = x 
        if y < temp_bb_TL_y:
            temp_bb_TL_y = y
        if x+w > temp_bb_BR_x:
            temp_bb_BR_x = x+w 
        if y+h > temp_bb_BR_y:
            temp_bb_BR_y = y+h
    ROI = frame[temp_bb_TL_y:temp_bb_BR_y, temp_bb_TL_x:temp_bb_BR_x]  ----> Crop out ROI

Algorithm 2: Identification
def IDENTIFICATION(ROI)
    if ROI exists:
        if size(ROI) > thresholdArea:
            vehicle_info, points, st_code=detectAndDecode(ROI)
        else:
            ROI = SuperResolution("ESPCN_x4.pb", ROI)