Scipy UnivariateSpline 错误：(m>k) 隐藏 m 失败：fpcurf0:m=3

如何解决Scipy UnivariateSpline 错误：(m>k) 隐藏 m 失败：fpcurf0:m=3

我正在尝试增加灰度图像的亮度。为此，我想创建一个样条曲线。但是当我尝试使用 scipy.interpolate.UnivariateSpline 时，它​​会引发错误。

追溯：

---------------------------------------------------------------------------
error                                     Traceback (most recent call last)
<ipython-input-38-289216dd01e1> in <module>
      8 x=[0,128,255]
      9 y=[0,190,255]
---> 10 myLUT=spline_to_lookup_table(x,y)

<ipython-input-38-289216dd01e1> in spline_to_lookup_table(spline_breaks,break_values)
      1 def spline_to_lookup_table(spline_breaks: list,break_values: list):
----> 2     spl = UnivariateSpline(spline_breaks,break_values)
      4     return spl(range(256))

~/anaconda3/envs/computer-vision/lib/python3.8/site-packages/scipy/interpolate/fitpack2.py in __init__(self,x,y,w,bBox,k,s,ext,check_finite)
    200 
    201         # _data == x,xb,xe,n,t,c,fp,fpint,nrdata,ier
--> 202         data = dfitpack.fpcurf0(x,w=w,xb=bBox[0],203                                 xe=bBox[1],s=s)
    204         if data[-1] == 1:

error: (m>k) Failed for hidden m: fpcurf0:m=3

源代码：

def spline_to_lookup_table(spline_breaks: list,break_values: list):
    spl = UnivariateSpline(spline_breaks,break_values)
    return spl(range(256))
x=[0,255]
y=[0,255]
myLUT=spline_to_lookup_table(x,y)
img_curved=cv2.LUT(img_gray,myLUT).astype(np.uint8)

解决方法

事实上，你不能用三个点拟合三次样条：即使是单个三次抛物线也有四个参数。