滑块匹配方法返回起始 x 坐标。

[kvii]

slide_match 方法内部调用 get_target 方法时也返回了 target_x。反正 target_y 最后也返回了，那 target_x 也顺便返回一下呗？

源码 2660 行是 target_x 初始化逻辑，2685 行是 target_y 返回逻辑。

ddddocr/ddddocr/__init__.py

Lines 2656 to 2686 in 491ce02

	def slide_match(self, target_bytes: bytes = None, background_bytes: bytes = None, simple_target: bool = False,
	flag: bool = False):
	if not simple_target:
	try:
	target, target_x, target_y = self.get_target(target_bytes)
	target = cv2.cvtColor(np.asarray(target), cv2.IMREAD_ANYCOLOR)
	except SystemError as e:
	# SystemError: tile cannot extend outside image
	if flag:
	raise e
	return self.slide_match(target_bytes=target_bytes, background_bytes=background_bytes,
	simple_target=True, flag=True)
	else:
	target = cv2.imdecode(np.frombuffer(target_bytes, np.uint8), cv2.IMREAD_ANYCOLOR)
	target_y = 0
	target_x = 0
	background = cv2.imdecode(np.frombuffer(background_bytes, np.uint8), cv2.IMREAD_ANYCOLOR)
	background = cv2.Canny(background, 100, 200)
	target = cv2.Canny(target, 100, 200)
	background = cv2.cvtColor(background, cv2.COLOR_GRAY2RGB)
	target = cv2.cvtColor(target, cv2.COLOR_GRAY2RGB)
	res = cv2.matchTemplate(background, target, cv2.TM_CCOEFF_NORMED)
	min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
	h, w = target.shape[:2]
	bottom_right = (max_loc[0] + w, max_loc[1] + h)
	return {"target_y": target_y,
	"target": [int(max_loc[0]), int(max_loc[1]), int(bottom_right[0]), int(bottom_right[1])]}

因为有的滑块是 png 格式的，这种滑块图片四周有一圈透明的像素。如果能拿到 target_x 的话，在计算滑动距离时就能够获得更精准的结果。所以 target_x 也是有必要返回的。