Question

我必须用大量样本（最多4e6个样本）绘制信号。我使用的是python 2.7，Qt 4.8和pyqtgraph 0.9.10。

我做了类似的事情：

生成信号并对其进行抽取（在单独的QThread中）
绘制信号并调用setDownsample（）和clipToView（）（pyqtgraph的方法）

问题是当QThread for decimate计算时，GUI会冻结。这是我的代码：

class TestDecimate(QDialog):

    def __init__(self, parent=None):
        super(TestDecimate, self).__init__(parent)

        self.decimate_thread = GenerateBigPlotThread()

        layout = QGridLayout()

        graph_widget = pg.GraphicsLayoutWidget()
        graph_layout = graph_widget.addLayout()

        self.p = graph_layout.addPlot(title="Plot 2", col=1, row=2)
        self.curve1 = self.p.plot([], pen=(255, 0, 0))
        self.curve2 = self.p.plot([], pen=(255, 255, 0))
        self.curve3 = self.p.plot([], pen=(255, 0, 255))
        self.curve4 = self.p.plot([], pen=(0, 0, 255))
        self.p.showGrid(True, True, 0.3)

        self.legend = pg.LegendItem(offset=(-10,10))
        self.legend.addItem(self.curve1, name="Test")
        self.legend.addItem(self.curve2, name="Test1")
        self.legend.addItem(self.curve3, name="Test2")
        self.legend.addItem(self.curve4, name="Test3")
        self.legend.setParentItem(self.p.getViewBox())

        self.start_pb = QPushButton("Generate")

        layout.addWidget(graph_widget, 1, 1)
        layout.addWidget(self.start_pb, 2, 1)

        self.setLayout(layout)

        self.start_pb.clicked.connect(self.start_thread_for_decimate)
        self.decimate_thread.newData.connect(self.set_data)

    def set_data(self, y, z, n, m):
        print "Plotting"
        self.curve1.setData(y)
        self.curve2.setData(z)
        self.curve3.setData(n)
        self.curve4.setData(m)

        self.p.setDownsampling(ds=True, auto=True, mode='peak')
        self.p.setClipToView(clip=True)
        self.p.getViewBox().enableAutoRange(axis=ViewBox.XYAxes)
        print "Plotting done"


    def start_thread_for_decimate(self):
        self.decimate_thread.start()

if __name__ == "__main__":
     app = QApplication(sys.argv)
     main = TestDecimate()
     main.show()
     sys.exit(app.exec_())

用于计算和抽取的线程：

class GenerateBigPlotThread(QThread):

    newData = QtCore.pyqtSignal("PyQt_PyObject","PyQt_PyObject","PyQt_PyObject","PyQt_PyObject")

    def __init__(self,*arg,**kwargs):
        super(GenerateBigPlotThread, self).__init__(*arg,**kwargs)

    def run(self):

        self.num_points = 4e6

        x = np.arange(self.num_points)
        y = np.sin(3.14159 * x * 10/self.num_points)
        z = np.cos(3.14159 * x * 10/self.num_points)
        n = np.sin(3.14159 * x * 5/self.num_points)
        m = np.cos(3.14159 * x * 5/self.num_points)
        print "Decimate"
        y = decimate(y, 4, ftype='fir')
        z = decimate(z, 4, ftype='fir')
        n = decimate(n, 4, ftype='fir')
        m = decimate(m, 4, ftype='fir')
        print "Done!"

        self.newData.emit(y, z, n, m)

启动线程时，查看问题的最佳方法是移动图例。此外，我尝试scipy.signal.resample（4e6样本到1e6样本）而不是抽取，并发生同样的问题。有人能告诉我这种行为的原因是什么？

Answer 1

问题在于scipy的版本。我有scipy 0.13.2，在我升级到0.17.0之后，问题似乎已经解决了。不过，我想知道为什么这样的事情一开始就有可能。

scypy.signal.decimate（）冻结了GUI

1 个答案: