PyQt5实现图像抠图交互式系统【环境搭建+GUI界面+功能实现】附完整代码

目录

 

1.  安装PyQt5

1.1 pip安装PyQt5

1.2 pip 安装 QtTools

1.3 配置PyCharm + QtDesigner

1.4 配置PyCharm+PyUic快捷菜单工具配置

1.5 验证安装是否成功

2、设计GUI界面

3、按钮功能实现

3.1 加载图像

3.2 轮廓查找

3.3 边缘检测

3.4 固定阈值抠图

3.5 自适应阈值抠图

3.6 HSV颜色抠图

 

---

 

1.  安装PyQt5

1.1 pip安装PyQt5

打开cmd窗口，使用命令 conda env list 查看当前虚拟环境，可以切换到你想要的虚拟环境进行下载，命令 activate 你的环境名字。

这里我就使用默认的环境base。使用命令 pip install PyQt5 安装PyQt5，若指定版本使用命令 pip install PyQt5==version。

pip install PyQt5 -i https://pypi.tuna.tsinghua.edu.cn/simple/

1.2 pip 安装 QtTools

注意，若PyQt5指定了版本，pyqt5-tools也要指定版本，否则pyqt5-tools安装为最新版本后，会将PyQt5也重新安装为最新版本。

pip install PyQt5-tools -i https://pypi.tuna.tsinghua.edu.cn/simple/

安装完毕后，可以使用命令 pip list查看已安装的第三方库：

1.3 配置PyCharm + QtDesigner

打开pycharm，菜单栏File -> Settings打开设置界面

进入Tools->External Tools->点击加号（+）

找到你的designer.exe所在位置，一般在你的虚拟环境下的 Lib\site-packages\ 下面。

Program:D:\anaconda\Lib\site-packages\qt5_applications\Qt\bin\designer.exe

Working directory: $FileDir$ 

 

1.4 配置PyCharm+PyUic快捷菜单工具配置

便于将Qt的ui文件转换成.py文件，配置参数如下：

Description: convert .ui to .py

Program:D:\anaconda\Scripts\pyuic5.exe（对应pyuic5.exe的路径）

Arguments: $FileName$ -o $FileNameWithoutExtension$.py

Working directory: $FileDir$ 

 1.5 验证安装是否成功

 

打开是这个界面就是安装成功了：

 

2、设计GUI界面

你可以在Qt Designer中选择组件来构建你想要的界面，然后再将其转换为py文件，再写一个main.py文件来给各个组件赋予功能。也可以直接写一个py文件来实现界面和功能的交互式操作。如下示例：

 

代码如下：

def init_ui(self):
    self.setWindowTitle("图像处理")
    self.setGeometry(100, 100, 800, 600)
    # 创建一个标签用于显示上传的图片
    self.uploaded_image_label = QLabel(self)
    self.uploaded_image_label.setAlignment(Qt.AlignCenter)
    # 创建五个标签用于显示抠图后的图片
    self.contours_image_label = QLabel(self)
    self.contours_image_label.setAlignment(Qt.AlignCenter)
    # 边缘检测
    self.edges_image_label = QLabel(self)
    self.edges_image_label.setAlignment(Qt.AlignCenter)
    # 固定阈值抠图
    self.segmented_image_label_fixed = QLabel(self)
    self.segmented_image_label_fixed.setAlignment(Qt.AlignCenter)
    # 自适应阈值抠图
    self.segmented_image_label_ada = QLabel(self)
    self.segmented_image_label_ada.setAlignment(Qt.AlignCenter)
    # HSV颜色范围抠图
    self.segmented_image_label_hsv = QLabel(self)
    self.segmented_image_label_hsv.setAlignment(Qt.AlignCenter)
    # 创建按钮
    self.load_button = QPushButton("选择图像", self)
    self.load_button.clicked.connect(self.load_image)
    self.contours_button = QPushButton("轮廓查找", self)
    self.contours_button.clicked.connect(self.display_contour)
    self.canny_button = QPushButton("Canny边缘检测", self)
    self.canny_button.clicked.connect(self.display_canny_edge)
    self.threshold_button = QPushButton("固定阈值抠图", self)
    self.threshold_button.clicked.connect(self.threshold_segmentation_dialog)
    self.adaptive_threshold_button = QPushButton("自适应阈值抠图", self)
    self.adaptive_threshold_button.clicked.connect(self.adaptive_threshold_segmentation_dialog)
    self.hsv_button = QPushButton("HSV颜色范围抠图", self)
    self.hsv_button.clicked.connect(self.hsv_segmentation_dialog)
    self.close_button = QPushButton("关闭", self)
    self.close_button.clicked.connect(self.close)
    self.spacer_button = QPushButton("帮助", self)
    self.spacer_button.setEnabled(False)  # 禁用按钮，使其看起来为空
    # 创建一个网格布局，并将按钮和标签添加到布局中
    layout = QGridLayout()
    layout.addWidget(self.load_button, 0, 0)
    layout.addWidget(self.contours_button, 0, 1)
    layout.addWidget(self.canny_button, 0, 2)
    layout.addWidget(self.spacer_button, 0, 3)
    layout.addWidget(self.threshold_button, 1, 0)
    layout.addWidget(self.adaptive_threshold_button, 1, 1)
    layout.addWidget(self.hsv_button, 1, 2)
    layout.addWidget(self.close_button, 1, 3)
    layout.addWidget(self.uploaded_image_label, 2, 0)
    layout.addWidget(self.contours_image_label, 2, 1)
    layout.addWidget(self.edges_image_label, 2, 2)
    layout.addWidget(self.segmented_image_label_fixed, 3, 0)
    layout.addWidget(self.segmented_image_label_ada, 3, 1)
    layout.addWidget(self.segmented_image_label_hsv, 3, 2)
    # 设置load_button的样式
    self.load_button.setStyleSheet("background-color: green; color: white; font-weight: bold;")
    # 设置close_button的样式
    self.close_button.setStyleSheet("background-color: orange; color: white; font-weight: bold;")
    self.spacer_button.setStyleSheet("background-color: blue; color: white; font-weight: bold;")
    # 设置其他按钮的样式
    button_style = "QPushButton { background-color: gray; color: white; font-weight: bold;}"
    self.contours_button.setStyleSheet(button_style)
    self.canny_button.setStyleSheet(button_style)
    self.threshold_button.setStyleSheet(button_style)
    self.adaptive_threshold_button.setStyleSheet(button_style)
    self.hsv_button.setStyleSheet(button_style)
    # 设置按钮的宽度
    self.load_button.setFixedWidth(200)
    self.contours_button.setFixedWidth(200)
    self.canny_button.setFixedWidth(200)
    self.threshold_button.setFixedWidth(200)
    self.adaptive_threshold_button.setFixedWidth(200)
    self.hsv_button.setFixedWidth(200)
    self.close_button.setFixedWidth(100)
    self.spacer_button.setFixedWidth(100)
    # 创建一个部件，将布局设置为部件的布局
    widget = QWidget()
    widget.setLayout(layout)
    # 设置部件为主窗口的中心部件
    self.setCentralWidget(widget)

 

3、按钮功能实现

• 文件处理模块：导入、导出数据文件和图像；
• 显示图像中间处理过程：轮廓查找、边缘检测；
• 抠图算法选择模块：固定阈值抠图、自适应阈值抠图、HSV颜色范围抠图；
• 系统模块：帮助和退出。

3.1 加载图像

代码如下：

def load_image(self):
    self.image_path = QFileDialog.getOpenFileName(self, "选择图像", "", "Images (*.png *.xpm *.jpg *.bmp)")[0]
    if self.image_path:
        pixmap = QPixmap(self.image_path)
        self.uploaded_image_label.setPixmap(
            pixmap.scaled(self.uploaded_image_label.size(),Qt.KeepAspectRatio, Qt.SmoothTransformation))

def display_image_segment(self,segmented_image_path,segment):
    if segmented_image_path:
        pixmap = QPixmap(segmented_image_path)
        segment.setPixmap(pixmap.scaled(segment.size(), Qt.KeepAspectRatio, Qt.SmoothTransformation))

 

3.2 轮廓查找

代码如下：

def perform_contour_detection(self):
    image = cv2.imread(self.image_path)
    gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    _, thresholded = cv2.threshold(gray_image, 128, 255, cv2.THRESH_BINARY)
    contours, _ = cv2.findContours(thresholded, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    contours_image = np.zeros_like(image)
    cv2.drawContours(contours_image, contours, -1, (255, 255, 255), 2)
    segmented_image_path_contour = "contours_image.png"
    cv2.imwrite(segmented_image_path_contour, contours_image)
self.display_image_segment(segmented_image_path_contour, self.contours_image_label)

 

3.3 边缘检测

代码如下：

def perform_canny_edge_detection(self):
    image = cv2.imread(self.image_path, cv2.IMREAD_GRAYSCALE)
    edges = cv2.Canny(image, 100, 200)
    segmented_image_path_edge = "edges_image.png"
    cv2.imwrite(segmented_image_path_edge, edges)
self.display_image_segment(segmented_image_path_edge, self.edges_image_label)

 

3.4 固定阈值抠图

代码如下：

def threshold_segmentation_dialog(self):
    threshold = QInputDialog.getInt(self, "固定阈值抠图", "阈值:", 128, 0, 255)
    image = cv2.imread(self.image_path, cv2.IMREAD_GRAYSCALE)
    _, thresholded = cv2.threshold(image, threshold[0], 255, cv2.THRESH_BINARY)
    segmented_image_path = "fixed_threshold.png"  # 保存抠图后的图像路径
    cv2.imwrite(segmented_image_path, thresholded)
    self.display_image_segment(segmented_image_path,self.segmented_image_label_fixed)

 

3.5 自适应阈值抠图

代码如下：

def adaptive_threshold_segmentation_dialog(self):
    image = cv2.imread(self.image_path, cv2.IMREAD_GRAYSCALE)
    thresholded = cv2.adaptiveThreshold(image, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 11, 2)
    segmented_image_path = "adaptive_threshold.png"  # 保存抠图后的图像路径
    cv2.imwrite(segmented_image_path, thresholded)
    self.display_image_segment(segmented_image_path,self.segmented_image_label_ada)

 

3.6 HSV颜色抠图

代码如下：

def hsv_segmentation_dialog(self):
    lower_hue, _ = QInputDialog.getInt(self, "HSV颜色范围抠图", "下限Hue:", 0, 0, 179)
    upper_hue, _ = QInputDialog.getInt(self, "HSV颜色范围抠图", "上限Hue:", 179, 0, 179)
    lower_saturation, _ = QInputDialog.getInt(self, "HSV颜色范围抠图", "下限Saturation:", 0, 0, 255)
    upper_saturation, _ = QInputDialog.getInt(self, "HSV颜色范围抠图", "上限Saturation:", 255, 0, 255)
    lower_value, _ = QInputDialog.getInt(self, "HSV颜色范围抠图", "下限Value:", 0, 0, 255)
    upper_value, _ = QInputDialog.getInt(self, "HSV颜色范围抠图", "上限Value:", 255, 0, 255)
    image = cv2.imread(self.image_path)
    image_hsv = cv2.cvtColor(image, cv2.COLOR_BGR2HSV)
    lower_range = np.array([lower_hue, lower_saturation, lower_value])
    upper_range = np.array([upper_hue, upper_saturation, upper_value])
    mask = cv2.inRange(image_hsv, lower_range, upper_range)
    thresholded = cv2.bitwise_and(image, image, mask=mask)
    segmented_image_path = "hsv_threshold.png"  # 保存抠图后的图像路径
    cv2.imwrite(segmented_image_path, thresholded)
    self.display_image_segment(segmented_image_path,self.segmented_image_label_hsv)

将各个按钮的功能代码实现后，点击运行，结果如下：

 

这只是一个示例，可以根据需要对算法进行优化，抠出理想的结果。完整代码上传到gitee，自取：https://gitee.com/ury-u/pytorch.git

（Cutout文件夹里面）

 

 

参考文章：

PyCharm安装配置PyQt5/QtDesigner/PyUic的超详细教程_pycharm安装pyqt-CSDN博客