C++的Release版怎么不让他优化删减变量

在 C 语言中，如果要禁用优化，可以使用编译器选项 "-O0"。例如，如果你使用 GCC 编译 C 程序，你可以使用命令 "gcc -O0 myprogram.c -o myprogram" 来编译不进行优化的可执行文件。 通常，优化是由编译器在编译时进行的，因此如果你想要禁用优化，你需要在编译时指定 "-O0" 选项。然而，如果你想要在运行时禁用优化，可能需要使用特定的函数或环境变量。你可以查看你使用的编译器的文档来了解更多信息。 请注意，禁用优化可能会导致程序的性能降低，因此通常不建议这样做。如果你在调试程序时需要禁用优化，可以在调试完成后再恢复优化。

将 C++ 代码转换为 Python 代码涉及一些语法和库函数的调整。以下是完整的 Python 版本的代码，保留了原有的逻辑和功能： ```python import cv2 import numpy as np import math # 定义常量 jifen = 5 daodianju = 3 feichezhanbi = 0.73 yuanhuanyuzhi = 40 yuanhuanjingdu = 10 kuanzeng = 90 kuanzeng2 = 70 CAMERA_H = 60 CAMERA_W = 94 OUT_H = 60 OUT_W = 94 # 全局变量 shiziji = 0 banmasign = 0 imgSour = None class Up: def __init__(self): self.dian = cv2.Point() self.que = 0 # 图片或视频二值化（动态阈值） def er_zhi(imgxiu): imgCvt = cv2.cvtColor(imgxiu, cv2.COLOR_BGR2GRAY) imgTemp = cv2.GaussianBlur(imgCvt, (5, 5), 0) _, imgThr = cv2.threshold(imgTemp, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU) cv2.imshow("二值化", imgThr) return imgThr # 画出赛道两侧线和中线 def che_dao_xian(imgEr): global shiziji, banmasign, imgSour imgCan = imgDil = None shizipan = 0 yuanhuanzuosign = 0 yuanhuanyousign = 0 yuanhuan = [] banmapan = 0 banmasign = 0 banmadi = [0, 0, 0] banmashang = [0, 0, 0] high = int(imgEr.shape[0] * feichezhanbi) # 获取每一列白色点数据 lieBais = [0] * imgEr.shape[1] for x in range(imgEr.shape[1]): for y in range(high - 1, -1, -1): if imgEr[y, x] != 0: lieBais[x] += 1 else: break if x > 0 and abs(lieBais[x] - lieBais[x - 1]) >= 60: banmasign += 1 if banmasign >= 6: banmapan = 1 # 左右寻最长白条 maxzuo = [0, 0] maxyou = [0, 0] for a in range(len(lieBais)): if lieBais[a] > maxzuo[1]: maxzuo[0] = a maxzuo[1] = lieBais[a] for a in range(len(lieBais) - 1, -1, -1): if lieBais[a] > maxyou[1]: maxyou[0] = a maxyou[1] = lieBais[a] # 找车道线 chushi = 0 chuju = 0 shiziZuox = 0 shiziZuoy = 0 shiziYoux = 0 zuoDao = [] youDao = [] zhongXian = [] for y in range(high - 1, high - maxzuo[1], -1): zuox = maxzuo[0] for x in range(maxzuo[0], -1, -1): if imgEr[y, x] != 0 and imgEr[y, x - 1] == 0: if len(zuoDao) >= 1: if abs(x - zuoDao[-1].dian.x) <= daodianju and abs(y - zuoDao[-1].dian.y) <= daodianju: zuoDao.append(Up()) zuoDao[-1].dian = cv2.Point(x, y) zuoDao[-1].que = 0 zuox = x if banmapan == 1: banmapan = 2 break else: zuox = x zuoDao.append(Up()) zuoDao[-1].dian = cv2.Point(x, y) zuoDao[-1].que = 0 break elif imgEr[y, x] != 0 and imgEr[y, x - 1] != 0 and imgEr[y, x - 2] != 0: shiziZuox += 1 if x - 3 == 0: break tempx = maxyou[0] - 4 for x in range(maxyou[0] - 4, imgEr.shape[1]): if imgEr[y, x] != 0 and imgEr[y, x + 1] == 0: if len(youDao) >= 1: if abs(x - youDao[-1].dian.x) <= daodianju and abs(y - youDao[-1].dian.y) <= daodianju: youDao.append(Up()) youDao[-1].dian = cv2.Point(x, y) youDao[-1].que = 0 tempx = x break else: tempx = x youDao.append(Up()) youDao[-1].dian = cv2.Point(x, y) youDao[-1].que = 0 break elif imgEr[y, x] != 0 and imgEr[y, x + 1] != 0 and imgEr[y, x + 2] != 0: shiziYoux += 1 if x + 3 == imgEr.shape[1] - 1: break if imgEr.shape[1] - tempx >= 80: if banmapan == 2 and imgEr[y, tempx + 62] != 0 and imgEr[y, tempx + 63] != 0 and imgEr.shape[0] - maxzuo[1] <= 80: banmadi[0] = zuox banmadi[1] = tempx banmadi[2] = y banmapan = 3 if chuju == 0: chuju = tempx - zuox if abs(shiziZuox - maxzuo[0]) <= 5 and abs(imgEr.shape[1] - maxyou[0] + 4 - shiziYoux) <= 5: shizipan = 1 # 判断圆环存在否，及其位置 if abs(len(zuoDao) - len(youDao)) >= yuanhuanyuzhi: ji1 = 0 ji2 = 0 signs = 0 # 左圆环 if len(zuoDao) < len(youDao): zuominx = 999999 for n in range(len(zuoDao)): if zuominx > zuoDao[n].dian.x: zuominx = zuoDao[n].dian.x for n in range(len(youDao)): signs = 0 for x in range(youDao[n].dian.x, max(zuominx - 5, 1), -1): if imgEr[youDao[n].dian.y, x] != 0 and imgEr[youDao[n].dian.y, x - 1] == 0: if yuanhuanyousign == 1 and ji1 >= 5: yuanhuanyousign = 2 yuanhuanweidian[1].x = youDao[n - 5].dian.x - kuanzeng yuanhuanweidian[1].y = youDao[n - 5].dian.y ji1 = 0 elif yuanhuanyousign == 3 and ji1 >= 5: yuanhuanyousign = 4 ji1 = 0 break signs = 1 ji1 += 1 ji2 = 0 if signs == 0: if yuanhuanyousign == 0 and ji2 >= 5: yuanhuanyousign = 1 yuanhuanweidian[0].x = youDao[n - 5].dian.x - kuanzeng yuanhuanweidian[0].y = youDao[n - 5].dian.y ji2 = 0 elif yuanhuanyousign == 2 and ji2 >= 5: yuanhuanyousign = 3 ji2 = 0 ji2 += 1 ji1 = 0 # 右圆环 else: youmaxx = 0 for n in range(len(youDao)): if youmaxx < youDao[n].dian.x: youmaxx = youDao[n].dian.x for n in range(len(zuoDao)): signs = 0 for x in range(zuoDao[n].dian.x, youmaxx): if imgEr[zuoDao[n].dian.y, x] != 0 and imgEr[zuoDao[n].dian.y, x + 1] == 0: if yuanhuanzuosign == 1 and ji1 >= yuanhuanjingdu: yuanhuanzuosign = 2 yuanhuanweidian[1].x = zuoDao[n - 5].dian.x + kuanzeng2 yuanhuanweidian[1].y = zuoDao[n - 5].dian.y ji1 = 0 elif yuanhuanzuosign == 3 and ji1 >= yuanhuanjingdu: yuanhuanzuosign = 4 ji1 = 0 break signs = 1 ji1 += 1 ji2 = 0 if signs == 0: if yuanhuanzuosign == 0 and ji2 >= yuanhuanjingdu: yuanhuanzuosign = 1 yuanhuanweidian[0].x = zuoDao[n - 5].dian.x + kuanzeng2 + 10 yuanhuanweidian[0].y = zuoDao[n - 5].dian.y ji2 = 0 elif yuanhuanzuosign == 2 and ji2 >= yuanhuanjingdu: yuanhuanzuosign = 3 ji2 = 0 ji2 += 1 ji1 = 0 # 十字判断 if shizipan: shiziji += 1 if shiziji >= 3: shiziji = 0 cv2.putText(imgSour, "CROSS", (10, 40), cv2.FONT_HERSHEY_COMPLEX, 0.7, (0, 0, 255), 2) # 斑马线判断 if banmapan == 3: cv2.putText(imgSour, "BANMA_WIRES", (10, 40), cv2.FONT_HERSHEY_COMPLEX, 0.7, (0, 0, 255), 2) cv2.line(imgSour, (banmadi[0], banmadi[2]), (banmadi[1], banmadi[2]), (255, 0, 255), 3.5) # 圆环判断 if yuanhuanzuosign == 4: cv2.putText(imgSour, "right circular", (10, 40), cv2.FONT_HERSHEY_COMPLEX, 0.7, (0, 0, 255), 2) for n in range(2): cv2.line(imgSour, (imgEr.shape[1] - 1, yuanhuanweidian[n].y), (yuanhuanweidian[n].x, yuanhuanweidian[n].y), (255, 0, 255), 3.5) elif yuanhuanyousign == 4: cv2.putText(imgSour, "left circular", (10, 40), cv2.FONT_HERSHEY_COMPLEX, 0.7, (0, 0, 255), 2) for n in range(2): cv2.line(imgSour, (0, yuanhuanweidian[n].y), (yuanhuanweidian[n].x, yuanhuanweidian[n].y), (255, 0, 255), 3.5) # 找中线 sizes = min(len(zuoDao), len(youDao)) bestZhongxian = [] niheZhongxian = [] kai = 0 zhongx = 0 zhongy = 0 for n in range(sizes): if zuoDao[n].que == 0 and youDao[n].que == 0: zhongx = (zuoDao[n].dian.x + youDao[n].dian.x) // 2 zhongy = (zuoDao[n].dian.y + youDao[n].dian.y) // 2 if kai == 0: zhongXian.append(Up()) zhongXian[-1].dian = cv2.Point(zhongx, zhongy) zhongXian[-1].que = 0 kai = 1 elif abs(zhongx - zhongXian[-1].dian.x) <= daodianju and abs(zhongy - zhongXian[-1].dian.y) >= 2: zhongXian.append(Up()) zhongXian[-1].dian = cv2.Point(zhongx, zhongy) zhongXian[-1].que = 0 # 最小二乘法拟合中线 for c in range(0, len(zhongXian), jifen): if c + jifen >= len(zhongXian): break aa = bb = xx = yy = xy = xfangs = 0 for n in range(c, c + jifen): xy += zhongXian[n].dian.x * zhongXian[n].dian.y xfangs += zhongXian[n].dian.x ** 2 xx += zhongXian[n].dian.x yy += zhongXian[n].dian.y xx /= jifen yy /= jifen bb = (xy - jifen * xx * yy) / (xfangs - jifen * xx ** 2) aa = yy - bb * xx zuomin = 999999 zuomax = 0 youmin = 999999 youmax = 0 for n in range(c, c + jifen): if zuomin > zuoDao[n].dian.x: zuomin = zuoDao[n].dian.x if zuomax < zuoDao[n].dian.x: zuomax = zuoDao[n].dian.x for n in range(c, c + jifen): if youmin > youDao[n].dian.x: youmin = youDao[n].dian.x if youmax < youDao[n].dian.x: youmax = youDao[n].dian.x for x in range(zuomin, youmax + 1): y = int(bb * x + aa) if y <= zhongXian[c].dian.y and y >= zhongXian[c + jifen].dian.y and abs(x - (zuomin + youmax) / 2) <= 3: niheZhongxian.append(cv2.Point(x, y)) # 拟合中线 for n in range(len(niheZhongxian) - 1): cv2.line(imgSour, niheZhongxian[n], niheZhongxian[n + 1], (0, 255, 0), 4) # 原始中线 for n in range(len(zhongXian) - 1): cv2.circle(imgSour, zhongXian[n].dian, 2, (0, 255, 0), -1) cv2.line(imgSour, zhongXian[n].dian, zhongXian[n + 1].dian, (0, 255, 0), 4) # 画出车道线 for n in range(len(zuoDao)): cv2.circle(imgSour, zuoDao[n].dian, 2, (255, 0, 0), -1) for n in range(len(youDao)): cv2.circle(imgSour, youDao[n].dian, 2, (0, 0, 255), -1) return imgSour # 逆透视 def ni_tou(): angle = 0.8 dep = 3.8 prop_j = 1 prop_i = 0 j_large = 1.6 i_abodon = 7 hight = 50 sin_a = math.sin(angle) cos_a = math.cos(angle) # 初始化摄像头坐标系 map_square = np.zeros((CAMERA_H, CAMERA_W, 2), dtype=np.float32) for i in range(CAMERA_H): for j in range(CAMERA_W): map_square[i, j, 0] = (CAMERA_H / 2 - i + 0.5) / 10 map_square[i, j, 1] = (j - CAMERA_W / 2 + 0.5) / 10 # 横向拉伸 for i in range(CAMERA_H): for j in range(CAMERA_W): map_square[i, j, 1] = map_square[i, j, 1] * (1 * (CAMERA_H - 1 - i) + (1 / prop_j) * i) / (CAMERA_H - 1) # 逆透视变换 for i in range(CAMERA_H): for j in range(CAMERA_W): xg = map_square[i, j, 1] yg = map_square[i, j, 0] y0 = (yg * dep + hight * cos_a * yg + hight * dep * sin_a) / (dep * cos_a - yg * sin_a) zt = -y0 * sin_a - hight * cos_a x0 = xg * (dep - zt) / dep map_square[i, j, 1] = x0 map_square[i, j, 0] = y0 # 横坐标缩放 prop_x = (OUT_W - 1) / (map_square[i_abodon, CAMERA_W - 1, 1] - map_square[i_abodon, 0, 1]) for i in range(CAMERA_H):