#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ 压缩相机客户端Python实现功能：连接服务器接收JPEG压缩视频流，解码并显示 """ import socket import cv2 import numpy as np import sys import time from PIL import Image import io def main(): print("Starting Compressed Camera Client...") if len(sys.argv) != 3: print(f"Usage: {sys.argv[0]} ") return 1 server_ip = sys.argv[1] port = int(sys.argv[2]) print(f"Configuration: Server IP={server_ip}, Port={port}") # 创建TCP套接字 print("Creating TCP socket...") sockfd = socket.socket(socket.AF_INET, socket.SOCK_STREAM) # 设置套接字超时 sockfd.settimeout(5.0) print("TCP socket created successfully") try: # 连接到服务器 print(f"Connecting to server {server_ip}:{port}...") sockfd.connect((server_ip, port)) print(f"Connected to server: {server_ip}:{port}") # 关闭超时，改为非阻塞模式 sockfd.setblocking(True) print("Socket set to blocking mode") # OpenCV窗口 print("Creating OpenCV window...") # 尝试使用不同的窗口参数 cv2.namedWindow("Compressed Camera Stream", cv2.WINDOW_GUI_NORMAL) # 设置固定的窗口大小 cv2.resizeWindow("Compressed Camera Stream", 640, 480) # 确保窗口大小是固定的 cv2.setWindowProperty("Compressed Camera Stream", cv2.WND_PROP_ASPECT_RATIO, cv2.WINDOW_KEEPRATIO) cv2.setWindowProperty("Compressed Camera Stream", cv2.WND_PROP_AUTOSIZE, cv2.WINDOW_AUTOSIZE) # 显示初始信息 info_frame = np.zeros((480, 640, 3), dtype=np.uint8) cv2.putText(info_frame, "Connecting to server...", (50, 240), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 255), 2) cv2.imshow("Compressed Camera Stream", info_frame) cv2.waitKey(100) frame_count = 0 buffer = b'' # 用于累积数据的缓冲区 total_bytes_received = 0 jpeg_found_count = 0 # 上次收到数据的时间 last_receive_time = time.time() while True: try: # 接收数据 data = sockfd.recv(8192) if not data: print("Server closed connection") return 0 last_receive_time = time.time() total_bytes_received += len(data) print(f"Received {len(data)} bytes, total: {total_bytes_received} bytes") # 将新数据添加到缓冲区 buffer += data # 仅保留最近的500KB数据，防止内存溢出 if len(buffer) > 500 * 1024: print(f"Buffer size exceeded 500KB, trimming to 500KB") buffer = buffer[-500 * 1024:] # 检查缓冲区中是否有JPEG标记 has_jpeg = b'\xff\xd8' in buffer print(f"Buffer contains JPEG SOI: {has_jpeg}, buffer size: {len(buffer)}") # JPEG文件格式标记 JPEG_START = b'\xff\xd8' # SOI (Start of Image) JPEG_END = b'\xff\xd9' # EOI (End of Image) # 查找所有完整的JPEG帧 while True: # 查找JPEG开始标记 start_idx = buffer.find(JPEG_START) if start_idx == -1: # 没有找到开始标记 print(f"No JPEG SOI found, buffer size: {len(buffer)}") # 清空大部分缓冲区，只保留末尾可能包含部分JPEG头的数据 if len(buffer) > 1024: buffer = buffer[-1024:] print(f"Trimmed buffer to {len(buffer)} bytes") break # 查找当前开始标记后的结束标记 end_idx = buffer.find(JPEG_END, start_idx + 2) if end_idx == -1: # 没有找到完整的结束标记 print(f"Found JPEG SOI but no EOI, buffer size from SOI: {len(buffer) - start_idx}") # 保留从开始标记到缓冲区末尾的数据 buffer = buffer[start_idx:] break # 提取完整的JPEG帧（包括结束标记） jpeg_frame = buffer[start_idx:end_idx + 2] jpeg_found_count += 1 print(f"Extracted JPEG frame {jpeg_found_count}, size: {len(jpeg_frame)} bytes") # 从缓冲区中移除已处理的JPEG帧 buffer = buffer[end_idx + 2:] print(f"Buffer remaining after extraction: {len(buffer)} bytes") # 只处理合理大小的JPEG帧（1KB到500KB） if 1024 <= len(jpeg_frame) <= 500 * 1024: try: # 使用PIL库来解码和显示图像 image_stream = io.BytesIO(jpeg_frame) pil_image = Image.open(image_stream) if pil_image is not None: # 解码成功，显示帧 frame_count += 1 print(f"Frame {frame_count} decoded successfully, size: {pil_image.width}x{pil_image.height}") # 转换为OpenCV格式 opencv_image = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR) # 检查图像是否有重复的模式 # 从截图来看，图像被分割成了3x3的网格 # 这可能是因为图像高度是正常的3倍 if opencv_image.shape[0] == 1440 and opencv_image.shape[1] == 640: # 高度是1440，是正常的3倍 # 只显示图像的前480行 fixed_image = opencv_image[:480, :, :] cv2.imshow("Compressed Camera Stream", fixed_image) print(f"Fixed frame by cropping to 640x480") elif opencv_image.shape[0] == 640 and opencv_image.shape[1] == 1920: # 宽度是1920，是正常的3倍 # 只显示图像的前640列 fixed_image = opencv_image[:, :640, :] cv2.imshow("Compressed Camera Stream", fixed_image) print(f"Fixed frame by cropping to 640x480") else: # 直接显示原始大小的图像 cv2.imshow("Compressed Camera Stream", opencv_image) # 只在显示新帧后调用一次waitKey，确保图像刷新 key = cv2.waitKey(1) if key == 27: print("ESC key pressed, exiting...") return 1 else: print(f"Failed to decode JPEG frame {jpeg_found_count}") except Exception as e: print(f"Error decoding JPEG frame {jpeg_found_count}: {e}") # 尝试使用OpenCV的方式解码 try: np_data = np.frombuffer(jpeg_frame, dtype=np.uint8) frame = cv2.imdecode(np_data, cv2.IMREAD_COLOR) if frame is not None: frame_count += 1 print(f"Frame {frame_count} decoded with OpenCV, size: {frame.shape[1]}x{frame.shape[0]}") # 检查是否是3x3网格 if frame.shape[0] > 480 and frame.shape[1] == 640: # 只显示前480行 cv2.imshow("Compressed Camera Stream", frame[:480, :, :]) else: cv2.imshow("Compressed Camera Stream", frame) key = cv2.waitKey(1) if key == 27: print("ESC key pressed, exiting...") return 1 except Exception as e2: print(f"Failed with OpenCV too: {e2}") else: print(f"Invalid JPEG frame size: {len(jpeg_frame)} bytes, skipped") except socket.timeout: print("Socket timeout, no data received") except Exception as e: print(f"Error in receive loop: {e}") # 检查连接是否超时 if time.time() - last_receive_time > 10: print("Connection timeout: No data received for 10 seconds") return 1 # 检查是否按下ESC键退出 key = cv2.waitKey(1) if key == 27: print("ESC key pressed, exiting...") break except socket.timeout as e: print(f"Connection timeout: {e}") return 1 except Exception as e: print(f"Error: {e}") return 1 finally: # 清理资源 print("Cleaning up resources...") sockfd.close() cv2.destroyAllWindows() print(f"Client exiting. Total bytes received: {total_bytes_received}, JPEG frames found: {jpeg_found_count}, Successful frames: {frame_count}") return 0 if __name__ == "__main__": sys.exit(main())