pyosv.freq.filters

  1import numpy as np
  2
  3def gaussian_filter(fft : np.ndarray, mx : int =0, my : int = 0, sx : int = 1, sy : int = 1, invert : bool = False) -> np.ndarray:
  4    '''
  5        Apply a 2D gaussian filter to the input 2D spectrum
  6        
  7        Prameters:
  8        ---------
  9            - fft : np.ndarray 
 10                2D spectrum to be filtered
 11            - mx : int 
 12                mean of gaussian x function (default : 0)
 13            - my : int
 14                mean of gaussian y function (default : 0)
 15            - sx : int
 16                standard deviation of gaussian x function (default : 1)
 17            = sy : int
 18                standard deviation of gaussian y function (default : 1)
 19            - invert : bool
 20                invert the distribution (default : False)
 21        
 22        Returns:
 23        --------
 24            - filt : np.ndarray
 25                filtered spectrum
 26        
 27        Usage:
 28        ------
 29        ```python
 30        import numpy as np
 31
 32        fft = np.array(
 33            [
 34            [0,   0, 1,   0,0],
 35            [0, 0.5, 1, 0.5,0],
 36            [1,   1, 1,   1,1],
 37            [0, 0.5, 1, 0.5,0],
 38            [0,   0, 1,   0,0],
 39            ]
 40
 41        )
 42
 43        filt = gaussian_filter(fft, mx=0, my=0, sx=1, sy=1, invert=False)
 44        ```
 45
 46        Output:
 47        -------
 48        ```
 49        array([  
 50            [0.        , 0.        , 0.09653235, 0.        , 0.        ],  
 51            [0.        , 0.061975  , 0.14045374, 0.061975  , 0.        ],  
 52            [0.09653235, 0.14045374, 0.15915494, 0.14045374, 0.09653235],  
 53            [0.        , 0.061975  , 0.14045374, 0.061975  , 0.        ],  
 54            [0.        , 0.        , 0.09653235, 0.        , 0.        ]])  
 55        ```
 56    '''
 57
 58    if len(fft.shape) < 2:
 59        raise Exception("Error: the shape of fft must be greater than 2")
 60    
 61    x = np.linspace(-1, 1, fft.shape[0])
 62    y = np.linspace(-1, 1, fft.shape[1])
 63    
 64    x, y = np.meshgrid(x,y)
 65
 66    fxy = 1. / (2. * np.pi * sx * sy) * np.exp(-((x - mx)**2. / (2. * sx**2.) + (y - my)**2. / (2. * sy**2.)))
 67
 68    if invert: fxy = 1-fxy
 69    
 70    filt = fft*fxy
 71    
 72    return filt
 73
 74
 75def lhp_filter(fft, radius=0.5, invert=False):
 76    '''
 77        Apply a low pass or high pass filter to the input 2D spectrum
 78        
 79        Parameters:
 80        -----------
 81            - fft : np.ndarray
 82                2D spectrum to be filtered
 83            - radius : int
 84                size of the filter, ammited range [0-1], will mask to zero fft values lower than radius (default : 0.5)
 85            - invert: bool
 86                invert the distribution (default : False)
 87
 88        Returns:
 89        --------
 90            - fft : np.ndarray
 91                filtered spectrum
 92
 93        
 94        Usage:
 95        ------
 96        ```python
 97        import numpy as np
 98
 99        fft = np.array(
100            [
101            [0,   0, 1,   0,0],
102            [0, 0.5, 1, 0.5,0],
103            [1,   1, 1,   1,1],
104            [0, 0.5, 1, 0.5,0],
105            [0,   0, 1,   0,0],
106            ]
107
108        )
109
110        filt = lhp_filter(fft, radius=0.8, invert=False)
111        ```
112
113        Output:
114        -------
115        ```
116        array([  
117            [0., 0., 1., 0., 0.],  
118            [0., 0., 0., 0., 0.],  
119            [1., 0., 0., 0., 1.],  
120            [0., 0., 0., 0., 0.],  
121            [0., 0., 1., 0., 0.]])  
122        ```
123    '''
124    
125    if len(fft.shape) < 2:
126        raise Exception("Error: the shape of fft must be greater than 2")
127    
128    x = np.linspace(-1, 1, fft.shape[0])
129    y = np.linspace(-1, 1, fft.shape[1])
130    x, y = np.meshgrid(x, y)
131
132    fxy = np.sqrt(x**2 + y**2)
133
134    if invert: fxy = 1-fxy
135
136    fft[fxy<radius] = 0
137    
138    return fft