def refine_gfimage(img, mask): #time-consuming-mask
# Draw BBox using gf
bboxes = findBBox(img, mask)
if bboxes is None:
return mask
box_img = np.zeros((img.shape), dtype=np.uint8)
#cv2.drawContours(box_img,[bboxes],-1,(255,255,255),1)
for bbox in bboxes:
bbox_ = resizeBBox(bbox, ratio=1.0)
cv2.drawContours(box_img, [bbox_], 0, (255, 255, 255), 1)
newmask = cv2.cvtColor(box_img, cv2.COLOR_BGR2GRAY)
#
#img_contours,contours,hierarchy = cv2.findContours(newmask, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)
contours, hierarchy = cv2.findContours(newmask, cv2.RETR_LIST,
cv2.CHAIN_APPROX_NONE)
img_contours = np.zeros((img.shape), dtype=np.uint8)
if len(contours) == 0:
return mask
for contour in contours:
random.shuffle(contour)
cv2.drawContours(img_contours, [contour], 0, (255, 255, 255), 1)
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 5))
newmask = cv2.morphologyEx(img_contours[:, :, 0], cv2.MORPH_CLOSE, kernel)
return newmask
def detect(self, origin, isdebug=False):
start = time.time()
# Default Size
h, w, c = origin.shape
size = 200.0
# Resize
img = cv2.resize(origin, (int(w * size / h), int(size)))
#showResult("img",img)
for case in switch(AnalyzeImageQuality.dayornight(img)):
if case('Day'):
# Extract Good Features
corners = refinedGoodFeatures(origin, img)
mask = checkFeatures(img, corners, isdebug)
closing = close(mask)
refined_gfmask = refine_gfimage(img, closing)
#showResult("refined_gfmask",refined_gfmask)
finalmasks = mkfinalmasks(img,
refined_gfmask,
isday=True,
isdebug=isdebug)
break
if case('Night'):
finalmasks = mkfinalmasks(img, None, isday=isdebug)
break
for colrindex, fmask in enumerate(finalmasks):
if (fmask > 0).sum() == 0:
continue
bboxes = mask2plates(img, fmask)
# Resize
if bboxes is not None:
bboxes = resizeBBoxes(bboxes, h / size)
rois = BBoxes2ROIs(origin, bboxes)
for i, roi in enumerate(rois):
confidence = self.licenplatevalidator.process(
roi, mode=colrs[colrindex], isdebug=isdebug)
print confidence
if confidence > 0.7:
#pts = self.licenplatevalidator.getRefinedROI()
#bbox = refineBBox(bboxes[i],pts)
bbox = resizeBBox(bboxes[i], ratio=0.9)
print("total elapsed time: " +
str(int((time.time() - start) * 1000) / 1000.0) +
"s")
return confidence, [bbox], [roi]
'''
# Check Result
if isdebug and bboxes is not None:
drawBBox(origin,bboxes,debug=True)
for i in range(len(rois)):
showResult("cropped",rois[i])
'''
return 0.0, None, None