def alphacv(self, img):
# This is an alternative computer vision algorithm proposed by Matthew
master = self.master
BLUR_FACTOR = master.values['BLUR']
TRACK_PT_NUM = master.values['PTS']
RADIUS = master.values['RADI']
ALPHA = master.values['A']
BETA = master.values['B']
GAMMA = 1 - ALPHA - BETA
THRESHOLD = master.values['THRS']
SAMPLESIZE = master.values['SIZE']
CERTAINTY = master.values['CERT']
# Do the image processing
# Generate grayscale image
grayimg = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# Blur image
blurred = findBlurred(grayimg, BLUR_FACTOR)
# The dynamic threshold calculation is embedded in algorithm
# TODO: Experiment to be done to determine the realibility of statistic
# Display necessary information on HUD
# Find tracking segments
interval = 15
pt_count = 10 #min(TRACK_PT_NUM, interval)
grayimg, pts = processing.get_good_pts(blurred, grayimg,
interval = interval, pt_count = pt_count)
master.cntframe = grayimg
master.trackingpts = pts
def captureAndSaveImage():
# Boot option: -c, captures image with recognition
stream = io.BytesIO()
CAMERA.resolution = (V_WIDTH, V_HEIGHT)
CAMERA.framerate = FRAMERATE
CAMERA.start_preview()
time.sleep(0.5)
CAMERA.capture(stream, format='jpeg')
data = np.fromstring(stream.getvalue(), dtype=np.uint8)
raw_img = cv2.imdecode(data, 0) # Returns a grayscale image
blurred = findBlurred(raw_img, CV_BLUR_FACTOR)
display, pointlst = processing.get_good_pts(blurred, raw_img,
interval=ALPHACV_INTERVAL, pt_count=ALPHACV_PT_COUNT,
skip=ALPHA_CV_ROW_SKIP, choose_thin=ALPHA_CV_CHOOSE_THIN)
# assert(len(pointlst) == 0): this is postcondition of alphacv
filename = 'captured/' + time.ctime().replace(' ', '_').replace(':', '_') \
+ '.jpg'
cv2.imwrite(filename, display)