图像的形态学操作

Question

我正在研究潜在指纹识别和分类系统。我在 colab 中对此进行了编码，其中代码运行良好，但是当我使用 UI 复制粘贴到 pycharm 时，它会输出黑色图像以用于侵蚀功能。

在我的代码中，图像增强后，我应用了腐蚀。下面显示的输出图像是我在colab上分别运行增强和腐蚀代码时得到的图像：

但是当我在 pycharm 上编码时：

norm, mask = __ridge_segment(img)  # normalise the image and find a ROI
orient = __ridge_orient(norm)  # compute orientation image

freq = __ridge_freq(norm, orient, mask)  # compute major frequency of ridges
binim = __ridge_filter(norm, freq, img, orient)  # filter the image using oriented gabor 
filter

binim1 = Image.fromarray(binim)

with BytesIO() as output:
    binim1.save(output, format="PNG")
    binim1 = output.getvalue()
    frame.update(data=binim1)
return binim

上面的代码在数组中输出带有布尔值的增强图像。

def Morph(out1,frame):

#out1 = np.array(out1)
out1=out1.copy()
out1 = np.array(out1).astype(np.uint8)
out1 = np.array(out1)
kernel = np.ones((2, 2), np.uint8)
kernel1 = np.ones((2, 2), np.uint8)
Erode = cv2.erode(out1, kernel, iterations=1)
Dilate = cv2.dilate(Erode, kernel1, iterations=1)
Morpho = Image.fromarray(Dilate)
sg.Popup('Ok clicked')

with BytesIO() as output:

    Morpho.save(output, format="PNG")
    Morpho = output.getvalue()
    frame.update(data=Morpho)


if __name__ == '__main__':
    sg.theme('Dark')
    sg.set_options(font=('Courier New', 11))

    w, h = size_of_image = (700, 600)

    layout_top = [
        [sg.InputText(enable_events=True, key='-FOLDER-'),
        sg.FolderBrowse('Browse', size=(7, 1), enable_events=True)],
        [sg.InputText(enable_events=True, key='-FILTER-'),
        sg.Button('Search', size=(7, 1))],[sg.Listbox([], size=(45, 10), enable_events=True,
            select_mode=sg.LISTBOX_SELECT_MODE_SINGLE, key='-LISTBOX-')],
    ]

    layout_bottom = [
    [sg.Button('Normalize', size=(20, 1),enable_events=True, key="normalize")],
    [sg.Button('Enhance', size=(20, 1),enable_events=True,key="Enhance")], 
[sg.Button('Morpholgy', size=(20, 1),enable_events=True,key='morph')],
    [sg.Button('Binarization', size=(20, 1))], [sg.Button('Thinning', size=(20, 1))],
    [sg.Button('Singular Points', size=(20, 1),enable_events=True, key= 'singular')], 
[sg.Button('Minutiae', size=(20, 1))],
    [sg.Button('Classify', size=(20, 1))], [sg.Button('Match', size=(20, 1))],
    ]
    layout_left = [
        [sg.Column(layout_top, pad=(0, 0))],
        [sg.Column(layout_bottom, pad=(0, 0))],
    ]
    layout_right = [[sg.Image(background_color='white', key='im')]]

    layout = [
        [sg.Column(layout_left), sg.Column(layout_right, pad=(0, 0), size=(w+15, h+15), 
background_color='lightblue', key='-COLUMN-')],

    ]

    window = sg.Window("PNG/GIF Viewer", layout, finalize=True)
    window['im'].Widget.pack(fill='both', expand=True)
    window['im'].Widget.master.pack(fill='both', expand=True)
    window['im'].Widget.master.master.pack(fill='both', expand=True)
    window['-COLUMN-'].Widget.pack_propagate(0)
    image=None
    while True:
        event, values = window.read()
        if event == sg.WINDOW_CLOSED:
            break
    # print(event, values)
        if event in ('-FOLDER-', '-FILTER-', 'Search'):
            update_listbox(window['-LISTBOX-'], values['-FOLDER-'],
                ('.png', '.gif'), values['-FILTER-'])
        elif event == '-LISTBOX-':
            lst = values['-LISTBOX-']
                if lst != []:
                image=update_image(window['im'], values['-LISTBOX-'][0])


        elif event == 'normalize':

            if image:

                image1=normalise(image, window['im'])
                #window['im'].update(data=image1)
                #image2=Enhance(image1,window['im'])
        if event=='Enhance':
           # if image1:
                image2 = Enhance(image1, window['im'])
                #window['im'].update(data=image2)

        if event == 'morph':
            image3 = Morph(image2, window['im'])

        if event == 'singular':
            # if image1:

            #image3 = Morph(image2, window['im'])
            #window['im'].update(data=image3)
                image3=singular_points(image1,window['im'])



    window.close()

上面的代码有 Morph 函数，它由腐蚀组成。

但是当增强的图像被侵蚀时，它会在屏幕上显示以下结果：

这是侵蚀后的表现：

有人能告诉我在代码中要修改什么，以便输出图像显示如上所示的侵蚀结果吗？

Answer 1

在函数中，我包含了out2 *= 255，它给出了结果。希望对和我有同样问题的人有所帮助。

def Morph(out1,frame):

    out1 = np.array(out1)
    out2=out1.copy()
    out2 = out2.astype(np.uint8)
    out2 *= 255              ### I added this line to make it work
    kernel = np.ones((2, 2), np.uint8)
    kernel1 = np.ones((2, 2), np.uint8)
    Erode = cv2.erode(out2, kernel, iterations=1)
    Dilate = cv2.dilate(Erode, kernel1, iterations=1)
    Morpho = Image.fromarray(Dilate)
    sg.Popup('Ok clicked')