In this article we will see how we can perform common image analysis using PyQtGraph module in Python. PyQtGraph is a graphics and user interface library for Python that provides functionality commonly required in designing and science applications. Its primary goals are to provide fast, interactive graphics for displaying data (plots, video, etc.) and second is to provide tools to aid in rapid application development (for example, property trees such as used in Qt Designer).
In order to install the PyQtGraph we use the command given below.
pip install pyqtgraph
Image analysis is the extraction of meaningful information from images; mainly from digital images by means of digital image processing techniques. Image analysis tasks can be as simple as reading bar coded tags or as sophisticated as identifying a person from their face.
In order to do this we have to do the following.
- Import the required libraries like pyqtgraph, pyqt5 and numpy.
- Create a main window class using pyqt5.
- Create a graphic window to add the widgets required to show the image analysis.
- Create two plot area and add a image item to it with the roi object to first plot area.
- Create an isocurve object and add it to the image item.
- Create a data for image and add it to the image item.
- Connect a update method to the roi object when the region is changed, inside the update method get the region and set it to the second plot area.
- Create a mouse move event and set the position, pixel value to the title according to the mouse position.
- Add this graph window to the main window layout with any additional widgets.
Below is the implementationÂ
Python3
# importing Qt widgetsfrom PyQt5.QtWidgets import *Â
# importing systemimport sysÂ
# importing numpy as npimport numpy as npÂ
# importing pyqtgraph as pgimport pyqtgraph as pgfrom PyQt5.QtGui import *from PyQt5.QtCore import *Â
class Window(QMainWindow):Â
    def __init__(self):        super().__init__()Â
        # setting title        self.setWindowTitle("PyQtGraph")Â
        # setting geometry        self.setGeometry(100, 100, 900, 550)Â
        # icon        icon = QIcon("skin.png")Â
        # setting icon to the window        self.setWindowIcon(icon)Â
        # calling method        self.UiComponents()Â
        # showing all the widgets        self.show()Â
    # method for components    def UiComponents(self):Â
        # creating a widget object        widget = QWidget()Â
        # text        text = "Image Analysis"Â
        # creating a label        label = QLabel(text)Â
        # setting minimum width        label.setMinimumWidth(130)Â
        # making label do word wrap        label.setWordWrap(True)Â
        # creating a graphic layout widgetÂ
        win = pg.GraphicsLayoutWidget()Â
        # plot area (ViewBox + axes) for displaying the image        p1 = win.addPlot(title="")Â
        # item for displaying image data        img = pg.ImageItem()Â
        # adding image to the plot area        p1.addItem(img)Â
        # Custom ROI for selecting an image region        roi = pg.ROI([-10, 14], [5, 5])        roi.addScaleHandle([0.5, 1], [0.5, 0.5])        roi.addScaleHandle([0, 0.5], [0.5, 0.5])Â
        # adding roi to the plot area        p1.addItem(roi)Â
        # setting z value to roi        # making sure ROI is drawn above image        roi.setZValue(10)Â
        # creating a Isocurve drawing on the image        iso = pg.IsocurveItem(level=1.2, pen='r')Â
        # setting parent as image        iso.setParentItem(img)Â
        # setting z axis value of isocurve        iso.setZValue(5)Â
        # Contrast/color control        hist = pg.HistogramLUTItem()Â
        # setting image to the control        hist.setImageItem(img)Â
        # adding control widget to the plot window        win.addItem(hist)Â
        # creating draggable line for setting isocurve level        isoLine = pg.InfiniteLine(angle=0, movable=True, pen='r')        hist.vb.addItem(isoLine)Â
        # making user interaction a little easier        hist.vb.setMouseEnabled(y=False)        isoLine.setValue(0.8)Â
        # bring iso line above contrast controls        isoLine.setZValue(1000)Â
        # going to next row of graphic window        win.nextRow()Â
        # another plot area for displaying ROI data        p2 = win.addPlot(colspan=2)Â
        # setting maximum height of plot area        p2.setMaximumHeight(250)Â
        # generating image data        data = np.random.normal(size=(200, 100))        data[20:80, 20:80] += 2.Â
        # setting gaussian filter to the data        data = pg.gaussianFilter(data, (3, 3))        data += np.random.normal(size=(200, 100)) * 0.1Â
        # setting data to the image        img.setImage(data)Â
        # setting level        hist.setLevels(data.min(), data.max())Â
        # build isocurves from smoothed data        iso.setData(pg.gaussianFilter(data, (2, 2)))Â
        # set position and scale of image        img.scale(0.2, 0.2)        img.translate(-50, 0)Â
        # zoom to fit image        p1.autoRange()Â
        # method for updating the plot        def updatePlot():Â
            # getting the selected region by the roi            selected = roi.getArrayRegion(data, img)Â
            # plot the selected region            p2.plot(selected.mean(axis=0), clear=True)Â
        # connecting the update plot method        # it get called when the region is changed        roi.sigRegionChanged.connect(updatePlot)Â
        # call the update plot method        updatePlot()Â
        # method for updating the isocurve        def updateIsocurve():            # setting iso level            iso.setLevel(isoLine.value())Â
        isoLine.sigDragged.connect(updateIsocurve)Â
        # method for image hover event        def imageHoverEvent(event):Â
            # showing the position, pixel, and value under the mouse cursor            # if cursor is not on the plot area            if event.isExit():                # set title as blank                p1.setTitle("")                returnÂ
            # getting cursor position            pos = event.pos()            i, j = pos.y(), pos.x()Â
            # pixel values            i = int(np.clip(i, 0, data.shape[0] - 1))            j = int(np.clip(j, 0, data.shape[1] - 1))Â
            # value of point            val = data[i, j]            ppos = img.mapToParent(pos)            x, y = ppos.x(), ppos.y()Â
            # setting plot title data            p1.setTitle(                "pos: (%0.1f, %0.1f) pixel: (%d, %d) value: %g" % (x, y, i, j, val))Â
        # Monkey-patch the image to use our custom hover function.        img.hoverEvent = imageHoverEventÂ
        # Creating a grid layout        layout = QGridLayout()Â
        # minimum width value of the label        label.setMinimumWidth(130)Â
        # setting this layout to the widget        widget.setLayout(layout)Â
        # adding label in the layout        layout.addWidget(label, 1, 0)Â
        # plot window goes on right side, spanning 3 rows        layout.addWidget(win, 0, 1, 3, 1)Â
        # setting this widget as central widget of the main window        self.setCentralWidget(widget)Â
# create pyqt5 appApp = QApplication(sys.argv)Â
# create the instance of our Windowwindow = Window()Â
# start the appsys.exit(App.exec()) |
Output:
