[Python] 纯文本查看 复制代码
# 更新StressShow命令
import numpy as np
import pyautogui
import copy
import cv2
from sklearn import cluster
class TuSe:
def __init__(self):
print('欢迎使用')
def GetCapture(self, stax, stay, endx, endy):
w = endx - stax
h = endy - stay
im = pyautogui.screenshot(region=(stax, stay, w, h))
# im = cv2.cvtColor(np.array(im), cv2.COLOR_BGR2RGB)
return np.array(im)
def FindPic(self, x1, y1, x2, y2, path, thd):
'''
找图
:param x1: 起点X
:param y1: 起点Y
:param x2: 终点X
:param y2: 终点Y
:param path: 图片路径
:param thd: 相似度
:return: 图片中心坐标
'''
img = self.GetCapture(x1, y1, x2, y2)
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
template = cv2.imread(path, 0)
th, tw = template.shape[::]
rv = cv2.matchTemplate(img, template, 1)
minVal, maxVal, minLoc, maxLoc = cv2.minMaxLoc(rv)
if 1 - minVal >= thd:
return minLoc[0] + tw / 2 + x1, minLoc[1] + th / 2 + y1
else:
return -1, -1
def Hex_to_Rgb(self, hex):
'''
十六进制转RGB
:param hex: 十六进制颜色值
:return: RGB
'''
return np.array(tuple(int(hex[i:i + 2], 16) for i in (0, 2, 4)))
def CmpColor(self, x, y, color, sim: float):
'''
比色
:param x: X坐标
:param y: Y坐标
:param color: 十六进制颜色，可以从大漠直接获取
:param sim: 相似偏移
:return: 真或加
'''
img = self.GetCapture(x - 1, y - 1, x + 1, y + 1)
img = np.array(img)
img = img[1][1]
color = self.Hex_to_Rgb(color)
res = np.absolute(color - img)
sim = int((1 - sim) * 255)
return True if np.amax(res) <= sim else False
def FindColor(self, x1, y1, x2, y2, des, sim: float):
'''
找色
:param x1: 起点X
:param y1: 起点Y
:param x2: 终点X
:param y2: 终点Y
:param des: 十六进制颜色，可以从大漠直接获取
:param sim: 相似偏移
:return:
'''
img = self.GetCapture(x1, y1, x2, y2)
img = np.array(img)
res = np.absolute(img - self.Hex_to_Rgb(des))
sim = int((1 - sim) * 255)
res = np.argwhere(np.all(res <= sim, axis=2))
res = res + (y1, x1)
return res[:, [1, 0]]
def GetColorNum(self, x1, y1, x2, y2, des, sim: float):
'''
获取颜色数量
:param x1: 起点X
:param y1: 起点Y
:param x2: 终点X
:param y2: 终点Y
:param des: 十六进制颜色，可以从大漠直接获取
:param sim: 相似偏移
:return:
'''
return len(self.FindColor(x1, y1, x2, y2, des, sim))
def FindMultColor(self, stax, stay, endx, endy, des):
'''
多点找色
:param stax:
:param stay:
:param endx:
:param endy:
:param des: 大漠获取到的多点找色数据，偏色必须写上
:return:
'''
w = endx - stax
h = endy - stay
img = pyautogui.screenshot(region=(stax, stay, w, h))
img = np.array(img)
rgby = []
ps = []
a = 0
firstXY = []
res = np.empty([0, 2])
for i in des.split(','):
rgb_y = i[-13:]
r = int(rgb_y[0:2], 16)
g = int(rgb_y[2:4], 16)
b = int(rgb_y[4:6], 16)
y = int(rgb_y[-2:])
rgby.append([r, g, b, y])
for i in range(1, len(des.split(','))):
ps.append([int(des.split(',')[i].split('|')[0]), int(des.split(',')[i].split('|')[1])])
for i in rgby:
result = np.logical_and(abs(img[:, :, 0:1] - i[0]) < i[3], abs(img[:, :, 1:2] - i[1]) < i[3],
abs(img[:, :, 2:3] - i[2]) < i[3])
results = np.argwhere(np.all(result == True, axis=2)).tolist()
if a == 0:
firstXY = copy.deepcopy(results)
else:
nextnextXY = copy.deepcopy(results)
for index in nextnextXY:
index[0] = int(index[0]) - ps[a - 1][1]
index[1] = int(index[1]) - ps[a - 1][0]
q = set([tuple(t) for t in firstXY])
w = set([tuple(t) for t in nextnextXY])
matched = np.array(list(q.intersection(w)))
res = np.append(res, matched, axis=0)
a += 1
unique, counts = np.unique(res, return_counts=True, axis=0)
index = np.argmax(counts)
re = unique[index] + (stay, stax)
if np.max(counts) == len(des.split(',')) - 1:
return np.flipud(re)
return np.array([-1, -1])
def FindPicEx(self, x1, y1, x2, y2, path, thd=0.9, MIN_MATCH_COUNT=8):
'''
全分辨率找图
:param x1:
:param y1:
:param x2:
:param y2:
:param path:
:param thd: 相似度
:param MIN_MATCH_COUNT: 特征点数量
:return:
'''
thd = thd - 0.2
template = cv2.imread(path, 0) # queryImage
# target = cv2.imread('target.jpg', 0) # trainImage
target = self.GetCapture(x1, y1, x2, y2)
target = cv2.cvtColor(target, cv2.COLOR_BGR2GRAY)
# Initiate SIFT detector创建sift检测器
sift = cv2.xfeatures2d.SIFT_create()
# find the keypoints and descriptors with SIFT
kp1, des1 = sift.detectAndCompute(template, None)
kp2, des2 = sift.detectAndCompute(target, None)
# 创建设置FLANN匹配
FLANN_INDEX_KDTREE = 0
index_params = dict(algorithm=FLANN_INDEX_KDTREE, trees=5)
search_params = dict(checks=50)
flann = cv2.FlannBasedMatcher(index_params, search_params)
matches = flann.knnMatch(des1, des2, k=2)
# store all the good matches as per Lowe's ratio test.
good = []
for m, n in matches:
if m.distance < thd * n.distance:
good.append(m)
if len(good) > MIN_MATCH_COUNT:
# 获取关键点的坐标
src_pts = np.float32([kp1[m.queryIdx].pt for m in good]).reshape(-1, 1, 2)
dst_pts = np.float32([kp2[m.trainIdx].pt for m in good]).reshape(-1, 1, 2)
# 计算变换矩阵和MASK
M, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0)
h, w = template.shape
# 使用得到的变换矩阵对原图像的四个角进行变换，获得在目标图像上对应的坐标
pts = np.float32([[0, 0], [0, h - 1], [w - 1, h - 1], [w - 1, 0]]).reshape(-1, 1, 2)
dst = cv2.perspectiveTransform(pts, M)
res = (dst[0] + dst[2]) / 2 # [[[ 39.11337 147.11575 ]] [[135.06624 255.12143 ]]
return int(res[0][0]) + x1, int(res[0][1]) + y1
else:
return -1, -1
def _FilterRec(self, res, loc):
""" 对同一对象的多个框按位置聚类后，按置信度选最大的一个进行保留。
:param res: 是 cv2.matchTemplate 返回值
:param loc: 是 cv2.np.argwhere(res>threshold) 返回值
:return: 返回保留的点的列表 pts
"""
model = cluster.AffinityPropagation(damping=0.5, max_iter=100, convergence_iter=10, preference=-50).fit(loc)
y_pred = model.labels_
pts = []
for i in set(y_pred):
argj = loc[y_pred == i]
argi = argj.T
pt = argj[np.argmax(res[tuple(argi)])]
pts.append(pt[::-1])
return np.array(pts)
def FindMultPic(self, x1, y1, x2, y2, path, thd):
'''
多目标找图
:param x1:
:param y1:
:param x2:
:param y2:
:param path:
:param thd: 相似度
:return:
'''
target = self.GetCapture(x1, y1, x2, y2)
target = cv2.cvtColor(target, cv2.COLOR_BGR2GRAY)
template = cv2.imread(path, 0)
w, h = template.shape[:2]
res = cv2.matchTemplate(target, template, cv2.TM_CCOEFF_NORMED)
loc = np.argwhere(res >= thd)
if len(loc):
resc = self._FilterRec(res, loc)
return resc + (h / 2 + x1, w / 2 + y1)
else:
return [[-1, -1]]
def FindPic_TM(self, x1, y1, x2, y2, path, thd):
'''
找透明图，透明色为黑色
:param x1: 起点X
:param y1: 起点Y
:param x2: 终点X
:param y2: 终点Y
:param path: 图片路径
:param thd: 相似度
:return: 图片中心坐标
'''
img = self.GetCapture(x1, y1, x2, y2)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
template = cv2.imread(path)
template2 = cv2.cvtColor(template, cv2.COLOR_BGR2GRAY)
ret, mask = cv2.threshold(template2, 20, 255, cv2.THRESH_BINARY)
th, tw = template.shape[:2]
rv = cv2.matchTemplate(img, template, 1, mask=mask)
minVal, maxVal, minLoc, maxLoc = cv2.minMaxLoc(rv)
if 1 - minVal >= thd:
return minLoc[0] + tw / 2 + x1, minLoc[1] + th / 2 + y1
else:
return -1, -1
def StressShow(self, stax, stay, endx, endy, des, type=0):
'''
保留选中颜色，其他为黑色，相似度根据偏色调整
:param stax:
:param stay:
:param endx:
:param endy:
:param des: 大漠的色彩描述
:param type: 0为原来颜色，1为白色
:return:
'''
# des = 'e81010-101010|f9ad08-000000'
dess = des.split('|')
des = [i[0:6] for i in dess]
des = [np.array(self.Hex_to_Rgb(d)) for d in des]
pds = [i[-6:] for i in dess]
pds = tuple(tuple(int(item[i:i + 2]) for i in range(0, len(item), 2)) for item in pds)
img = self.GetCapture(stax, stay, endx, endy)
mask = np.zeros(img.shape[:2], dtype=np.bool_)
for i, color in enumerate(des):
mask += np.all(np.abs(img - color) <= pds[i], axis=-1)
new_img = np.where(mask[..., None], [255, 255, 255], [0, 0, 0]) if type else np.where(mask[..., None], img,
[0, 0,
0]) # 修改这里，将选中的颜色设为白色
img_converted = cv2.convertScaleAbs(new_img)
img_converted = cv2.cvtColor(np.array(img_converted), cv2.COLOR_BGR2RGB)
return img_converted
a = TuSe()
b = a.StressShow(0, 0, 1920, 1080, 'e81010-101010|36659e-101010', 0)
cv2.imshow('13', b)
cv2.waitKey(0)
cv2.destroyAllWindows()
[复制链接]
发表于 2023-11-2 23:23
|
发表于 2023-11-3 07:43
|楼主
收藏
淘帖
有用
分享到朋友圈
