完善V1.4踢乒乓球版本

2026-01-20 22:15:03 +08:00
1 parent d14fe6c37b
commit 72455a5e76
@@ -0,0 +1,140 @@
+from maix import image, app, time
+
+# 橘黄色乒乓球的颜色阈值配置（HSV色域）
+# yxl
+thresholds = [
+    [73, 93, 7, 27, 13, 33],
+    [40, 65, 30, 60, 40, 70],
+    [39, 59, 41, 61, 48, 68],
+]
+
+# mgw
+thresholds = [
+    [60, 85, 10, 35, 60, 90],
+    [40, 65, 20, 50, 60, 90],
+    [30, 60, 30, 60, 40, 70],
+]
+
+# 丢失目标的阈值配置
+LOST_FRAME_THRESHOLD = 20  # 连续x帧没识别到才判定为真正丢失
+
+# 全局参数初始化
+kick_status = ""
+
+# ------------------------------
+# 兵乓球追踪
+# ------------------------------
+def select_largest_blob(blobs):
+    """
+    筛选并选取最大的有效blob（替换原融合逻辑）
+    1. 筛除宽/高小于10px的小blob（过滤噪声）
+    2. 从剩余blob中选取面积最大的单个blob
+    :param blobs: 识别到的所有blob列表
+    :return: 最大blob的矩形参数 (x, y, width, height) 和中心点 (cx, cy)；未识别到则返回None, None
+    """
+    if not blobs:
+        return None, (-1,-1)
+    
+    # 步骤1：过滤掉宽/高小于10px的无效blob
+    valid_blobs = []
+    for b in blobs:
+        x, y, w, h = b[0], b[1], b[2], b[3]
+        if w >= 10 and h >= 10:  # 筛除边距（宽/高）小于10px的blob
+            valid_blobs.append(b)
+    
+    # 过滤后无有效blob，返回None
+    if not valid_blobs:
+        return None, (-1,-1)
+    
+    # 步骤2：选取面积（w*h）最大的blob
+    largest_blob = max(valid_blobs, key=lambda b: b[2] * b[3])
+    
+    # 提取最大blob的矩形参数和中心点
+    x, y, w, h = largest_blob[0], largest_blob[1], largest_blob[2], largest_blob[3]
+    cx = x + w // 2
+    cy = y + h // 2
+    
+    return (x, y, w, h), (cx, cy)
+
+def calculate_deviation_ball(center, image_center):
+    """
+    优化：仅计算偏移量，不直接发送指令（将指令发送逻辑移到主循环统一处理）
+    :return: (deviation_x, deviation_y) 或 None（无有效目标）
+    """
+    center_x, center_y = center
+
+    if center_x == -1 and center_y == -1 :
+        return None  # 无有效目标
+    
+    # 计算x/y方向的偏移偏差（目标中心 - 图像中心）
+    deviation_x = center_x - image_center[0]
+    deviation_y = center_y - image_center[1]
+    
+    # 仅当偏移超过阈值时返回有效偏移，否则返回None（避免微小抖动）
+    if abs(deviation_x) > 10 or abs(deviation_y) > 10 :
+        return (deviation_x, deviation_y)
+    else:
+        return None
+
+
+def trace_ball_main(resources):
+
+    # 声明使用全局变量（两个参数都需要更新）
+    global kick_status
+
+    detector = resources["detector"]
+    cam = resources["cam"]
+    disp = resources["disp"]
+    serial_dev = resources["serial_dev"]
+    img_width = resources["img_width"]
+    img_height = resources["img_height"]
+    image_center = resources["image_center"]
+
+    last_valid_deviation = None  # 记录最后一次有效偏移量
+    lost_frames = 0              # 丢失帧数计数
+
+    while not app.need_exit():
+
+        img = cam.read().flip(image.FlipDir.Y)
+        bcmd = ""
+        
+        # 识别色块（像素阈值设为10，过滤小噪声）
+        blobs = img.find_blobs(thresholds, pixels_threshold=10)
+        
+        # 筛选并选取最大的有效blob
+        largest_rect, target_center = select_largest_blob(blobs)
+        
+        # 计算当前帧的偏移量
+        current_deviation = calculate_deviation_ball(target_center, image_center)
+
+        # -------------------------- 核心优化逻辑 --------------------------
+        if current_deviation is not None:
+            # 1. 识别到有效目标：重置丢失计数，更新最后一次有效偏移
+            lost_frames = 0
+            last_valid_deviation = current_deviation
+            dx, dy = current_deviation
+            bcmd = f"DX:{dx},DY:{dy};"
+        else:
+            # 2. 未识别到目标：累计丢失帧数
+            lost_frames += 1
+            if lost_frames < LOST_FRAME_THRESHOLD and last_valid_deviation is not None:
+                dx, dy = last_valid_deviation
+                bcmd = f"DX:{dx},DY:{dy};"
+            else:
+                # 2.2 丢失帧数达阈值：发送9999，重置最后有效偏移
+                bcmd = "DX:9999,DY:9999;"
+                last_valid_deviation = None
+        
+        # 绘制识别结果
+        if largest_rect:
+            x, y, w, h = largest_rect
+            img.draw_rect(x, y, w, h, image.COLOR_GREEN, 2)
+            cx, cy = target_center
+            img.draw_circle(cx, cy, 3, image.COLOR_RED, -1)
+
+
+        serial_dev.write_str(bcmd)
+        print(bcmd)
+
+        disp.show(img)
+        time.sleep_ms(1)