Module toolbox

Functions

def clean_tomo_data(df)

Function to select and sort relevant data from the output of porosity detection plugin.

The data in question are (x,y,z) coordinates of defect center of mass, defect volume and sphericity factor.

Args

df : Dataframe

Pandas dataframe containing porosity data.

Returns

Dataframe

Pandas df object.

Expand source code

def clean_tomo_data(df):
    """Function to select and sort relevant data from the output of porosity
    detection plugin.
    
    The data in question are (x,y,z) coordinates of defect center of mass, defect
    volume and sphericity factor.

    Args:
        df (Dataframe): Pandas dataframe containing porosity data.

    Returns:
        Dataframe : Pandas df object.
    """
    df = df[['xg', 'yg', 'zg', 'volpix', 'volmarch', 'sphericity']]
    df = df.sort_values('zg', ascending=False)
    df = df[df["volpix"] != df["volpix"].max()]
    return df

def compute_indicators_porosity_level(df, vol_layer)

Function to compute a porosity severity score based on the four metrics :

1 - Frequency of occurrence or number of discrete pores in a layer.

2 - The proportion of volume in a layer affected by porosity mu2.

3 - The average distance between a pair of pores mu3.

4 - The above three metrics are combined into a single metric called the normalized porosity level μ.

Args

df : Dataframe

The path of ImageJ installation.

vol_layer : float

The volume of part layer.

Returns

Dataframe

Pandas dataframe containing the four mu values for each

porosity.

Expand source code

def compute_indicators_porosity_level(df, vol_layer):
    """Function to compute a porosity severity score based on the four metrics :


    1 - Frequency of occurrence or number of discrete pores in a layer.
    
    2 - The proportion of volume in a layer affected by porosity mu2.
    
    3 - The average distance between a pair of pores mu3.
    
    4 - The above three metrics are combined into a single metric called the
    normalized porosity level μ.

    Args:
        df (Dataframe): The path of ImageJ installation.
        vol_layer (float): The volume of part layer.

    Returns:
        Dataframe : Pandas dataframe containing the four mu values for each
        porosity.

    """
    # occurrence frequency
    mu1 = pd.DataFrame(df.pivot_table(index=['numLayer'], aggfunc='size')
                      ).rename(columns={0: 'mu1'})
    mu1_norm = (mu1 - mu1.mean()) / mu1.std()

    # propotion of affected zone
    mu2 = pd.DataFrame(df.groupby('numLayer')['volmarch'].sum().div(vol_layer)
                      ).rename(columns={'volmarch': 'mu2'})
    mu2_norm = (mu2 - mu2.mean()) / mu2.std()

    # average distance between pores
    mu1_serie = df.pivot_table(index=['numLayer'], aggfunc='size')
    mu3 = pd.DataFrame(
        df.groupby('numLayer')[['xg',
                                'yg']].apply(lambda x: dist(x)).div(mu1_serie)
    ).rename(columns={0: 'mu3'})
    mu3_norm = (mu3 - mu3.mean()) / mu3.std()

    # porosity level
    mu = pd.DataFrame(
        mu2_norm['mu2'].mul(mu1_norm['mu1']).div(mu3_norm['mu3'])
    ).rename(columns={0: 'mu'})

    # normalised porosity level
    normalised_mu = (mu - mu.min()) / (mu.max() - mu.min())
    normalised_mu_V2 = (mu - mu.mean()) / mu.std()
    new_df = pd.concat([mu1, mu2, mu3, normalised_mu], axis=1)
    return new_df

def dist(df)

Expand source code

def dist(df):
    return sum(pdist(df.values, 'euclid'))

def label(IMAGEJ_PATH, IN_DIR, OUT_DIR, TOMO_LABELS_PATH)

Script Function to automate the labeling process using ImageJ plugin. It takes in the paths of local ImageJ installation which must include the plugin and of input data in form of Tif image stack.

The user must also specify the paths for the output which includes processed images and defect data in form of csv.

Args

IMAGEJ_PATH : string

The path of ImageJ installation.

IN_DIR : string

The path of input images.

OUT_DIR : string

The path to which output images will be written.

TOMO_LABELS_PATH : string

The path to which the data will be written.

Expand source code

def label(IMAGEJ_PATH, IN_DIR, OUT_DIR, TOMO_LABELS_PATH):
    """Script Function to automate the labeling process using ImageJ plugin.
    It takes in the paths of local ImageJ installation which must include the plugin
    and of input data in form of Tif image stack.
    
    The user must also specify the paths for the output which includes processed
    images and defect data in form of csv.

    Args:
        IMAGEJ_PATH (string): The path of ImageJ installation.
        IN_DIR (string): The path of input images.
        OUT_DIR (string): The path to which output images will be written.
        TOMO_LABELS_PATH (string): The path to which the data will be written.

    """
    ij = imagej.init(IMAGEJ_PATH, headless=False)
    tif_paths = []
    listOfFiles = os.listdir(IN_DIR)
    pattern = "*.tif"

    for entry in listOfFiles:
        if fnmatch.fnmatch(entry, pattern) and not fnmatch.fnmatch(entry, ".*"):
            print(entry)
            # selects tif files and appends them to a list
            tif_paths.append(entry)

    for index, tif in enumerate(tif_paths):
        # iterate over tif files
        # etval, mats = cv.imreadmulti(tif, flags=cv.IMREAD_GRAYSCALE)
        mats = tifio.imread(tif)
        assert len(mats) != 0, "No images were loaded"
        mats_th = []
        tif_processed = os.path.join(
            OUT_DIR,
            tif[:len(tif) - 4] + "_processed_" + str(index + 1) + ".tif"
        )
        for idx, mat in enumerate(mats):
            # iterates over all slices in each tif files
            mat = cv.fastNlMeansDenoising(
                mat, h=10, templateWindowSize=7, searchWindowSize=21
            )  # denoise
            _, gray_th = cv.threshold(
                mat, 0, 255, cv.THRESH_BINARY + cv.THRESH_OTSU
            )  # threshold
            # gray_th = imutils.rotate(gray_th, angles[index])  # rotate
            mats_th.append(gray_th)
        cv.imwritemulti(tif_processed, mats_th)

    for index, tif in enumerate(os.listdir(OUT_DIR)):

        ij_img = ij.io().open(OUT_DIR + '/' + tif)
        ij.ui().show('image', ij_img)

        plugin = 'Labeling'

        args = {'color': 0, 'minimum': 0, '3d': 6}
        ij.py.run_plugin(plugin, args)

        tif = os.path.splitext(tif)[0]
        plugin = 'Parameters'
        args = {'save': '/' + TOMO_LABELS_PATH + "/Labels_" + tif + ".dat"}
        ij.py.run_plugin(plugin, args)

        ij.dispose()

def load_tomo_data(file)

Function to load data into a pandas dataframe.

Args

file : string

The path to the csv dataset.

Returns

Dataframe

Pandas df object.

Expand source code

def load_tomo_data(file):
    """Function to load data into a pandas dataframe.

    Args:
        file (string): The path to the csv dataset.

    Returns:
        Dataframe : Pandas df object.
    """
    # filename = re.match(file_number)
    # csv_path = os.path.join(TOMO_LABELS_PATH, filename)
    df = pd.read_csv(file, delimiter=" ")
    return df