OpenCV Series : Target Box Outline Border


角点

c 复制代码
P1      [0]     (255, 000, 000)
P2      [1]     (000, 255, 000)
P3      [2]     (000, 000, 255)
P4      [3]     (000, 000, 000)


垂直矩形框

c 复制代码
	rect = cv2.minAreaRect(cnt)
	targetColor = roi_color
	targetThickness = 1
	targetColor = (255, 255, 255)
	if lineVerbose:
		if True:
		   cv2.line(photo,  (x, y), (x+w//4, y), targetColor, targetThickness)
		   cv2.line(photo,  (x, y), (x, y+h//4), targetColor, targetThickness)
		
		   cv2.line(photo,  (x+3*w//4, y), (x+w, y), targetColor, targetThickness)
		   cv2.line(photo,  (x+w, y), (x+w, y+h//4), targetColor, targetThickness)
		
		   cv2.line(photo,  (x, y+h), (x+w//4, y+h), targetColor, targetThickness)
		   cv2.line(photo,  (x, y+h), (x, y+3*h//4), targetColor, targetThickness)
		
		   cv2.line(photo,  (x+w, y+h), (x+3*w//4, y+h), targetColor, targetThickness)
		   cv2.line(photo,  (x+w, y+h), (x+w, y+3*h//4), targetColor, targetThickness)
		
		   crossLength = 15
		   cv2.line(photo,  (x+w//2-crossLength, y+h//2), (x+w//2+crossLength, y+h//2), targetColor, targetThickness)
		   cv2.line(photo,  (x+w//2, y+h//2-crossLength), (x+w//2, y+h//2+crossLength), targetColor, targetThickness)



倾斜矩形框

c 复制代码
def lineByAngle(photo, p1, p2, length, color, thickness):
    slope = 0 + (p2[1] - p1[1]) / (p2[0] - p1[0])
    theta = np.arctan(slope)
    degree = int(theta * 57.29577) % 360

    length = getDistance(p1, p2) // 4

    if (p2[0] < p1[0]):
        pp = (
            int(p1[0] - length * np.cos(theta)),
            int(p1[1] - length * np.sin(theta))
        )
    else:
        pp = (
            int(p1[0] + length * np.cos(theta)),
            int(p1[1] + length * np.sin(theta))
        )
    cv2.line(photo,  p1, pp, color, thickness)


几何关键点

c 复制代码
	if True:
	    cv2.circle(photo, p1, 1, roi_red, 5)
	    cv2.circle(photo, p2, 1, roi_red, 5)
	    cv2.circle(photo, p3, 1, roi_red, 5)
	    cv2.circle(photo, p4, 1, roi_red, 5)
	
	    center = (int((p1[0] + p3[0]) / 2), int((p2[1] + p4[1]) / 2))
	    cv2.circle(photo, center, 1, roi_green, 5)
c 复制代码
def slopeAngle(p1, p2):
    slope  = (p2[1] - p1[1]) / (p2[0] - p1[0])
    theta  = np.arctan(slope)
    degree = int(theta * 57.29577) % 360
    return theta

def lineByPoint(photo, pp, theta, length, color, thickness):
    if True:
        p1 = (
            int(pp[0] - length * math.cos(theta)),
            int(pp[1] - length * math.sin(theta))
        )
c 复制代码
     center = (int((p1[0] + p3[0]) / 2), int((p2[1] + p4[1]) / 2))
     if not True:
         cv2.circle(photo, p1, 1, roi_blue,  5)
         cv2.circle(photo, p2, 1, roi_green, 5)
         cv2.circle(photo, p3, 1, roi_red,   5)
         cv2.circle(photo, p4, 1, roi_black, 5)
         cv2.circle(photo, center, 1, roi_green, 5)
     if True:
         crossLength = 15
         theta = slopeAngle(p1, p2)
         lineByPoint(photo, center, theta, crossLength, targetColor, targetThickness)
         theta = slopeAngle(p2, p3)
         lineByPoint(photo, center, theta, crossLength, targetColor, targetThickness)


Automatic Mask

c 复制代码
if True:
    matrix = np.array(cnt)
    for i in range(len(matrix)):
        matrix[i][0][0] = matrix[i][0][0] - x
        matrix[i][0][1] = matrix[i][0][1] - y
    mask = np.zeros(characteristic.shape[:2], dtype=np.uint8)
    mask = cv2.fillPoly(mask, [matrix], 255)
    cv2.imshow('Mask', mask)

    image = cv2.bitwise_and(characteristic, characteristic, mask = mask)
    image = stretch(image)
    cv2.imshow('Masked', image)
相关推荐
出门吃三碗饭几秒前
CARLA: 如何在 CARLA 中回放自动驾驶场景
人工智能·机器学习·自动驾驶
Axis tech2 分钟前
第二届人形机器人半程马拉松即将于4月开赛,对比去年技术进步有哪些?
人工智能·机器人
志栋智能2 分钟前
超自动化巡检,如何成为业务稳定的“压舱石”?
大数据·运维·网络·人工智能·自动化
lifallen4 分钟前
从零推导一个现代 ReAct Agent框架
人工智能·算法·语言模型
我的offer在哪里4 分钟前
腾讯 Ardot 深度博客:AI 重构 UI/UX 全链路,从 “描述即界面” 到设计工业化的腾讯范式
人工智能·ui·重构
AEIC学术交流中心4 分钟前
【快速EI检索 | IEEE出版】第六届信号图像处理与通信国际学术会议(ICSIPC 2026)
图像处理·人工智能
康世行6 分钟前
IDEA集成AI辅助工具推荐(好用不卡顿)
java·人工智能·intellij-idea
柯儿的天空6 分钟前
【OpenClaw 全面解析:从零到精通】第007篇:流量枢纽——OpenClaw Gateway 网关深度解析
人工智能·gpt·ai作画·gateway·aigc·ai编程·ai写作
人道领域7 分钟前
2026年Q1大模型深度复盘:OpenAI,Gemini2.0,字节跳动,与“多模态Agent”元年
人工智能·ai·google·chatgpt·gemini
前端摸鱼匠8 分钟前
大模型面试题1:简述大模型(LLM)的定义,与传统NLP模型的核心区别是什么?
人工智能·ai·语言模型·自然语言处理·面试·职场和发展