我想处理彩色图像并仅保留非黑色像素。我写了一些代码，但它似乎没有做我认为它正在做的事情：
super@DESKTOP MINGW64 ~/OneDrive/Documents/kibana (master)
$ yarn kbn bootstrap
yarn run v1.6.0
$ node scripts/kbn bootstrap
Running [bootstrap] command from [C:\Users\super\OneDrive\Documents\kibana]:
Found [16] projects:
kibana
├── packages
│ ├── eslint-config-kibana (@elastic/eslint-config-kibana)
│ ├── eslint-plugin-kibana-custom (@elastic/eslint-plugin-kibana-custom)
│ ├── kbn-babel-preset (@kbn/babel-preset)
│ ├── kbn-datemath (@kbn/datemath)
│ ├── kbn-dev-utils (@kbn/dev-utils)
│ ├── kbn-es (@kbn/es)
│ ├── kbn-eslint-import-resolver-kibana (@kbn/eslint-import-resolver-kibana)
│ ├── kbn-eslint-plugin-license-header (@kbn/eslint-plugin-license-header)
│ ├── kbn-plugin-generator (@kbn/plugin-generator)
│ ├── kbn-plugin-helpers (@kbn/plugin-helpers)
│ ├── kbn-pm (@kbn/pm)
│ ├── kbn-system-loader (@kbn/system-loader)
│ ├── kbn-test-subj-selector (@kbn/test-subj-selector)
│ └── kbn-ui-framework (@kbn/ui-framework)
└── x-pack
Running installs in topological order:
Installing dependencies in [@kbn/babel-preset]:
[bootstrap] failed:
Error: Command failed: yarn install --non-interactive --mutex file
at makeError (C:\Users\super\OneDrive\Documents\kibana\packages\kbn-pm\dist\
index.js:38964:9)
at Promise.all.then.arr (C:\Users\super\OneDrive\Documents\kibana\packages\k
bn-pm\dist\index.js:39069:16)
at <anonymous>
at process._tickCallback (internal/process/next_tick.js:188:7)
error Command failed with exit code 1.
info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.
黑色像素为[0,0,0]，import cv2
import numpy as np
import tensorly as tl
def eq_svtn(T, b, M, Omega):
X = 0
for i in range(3):
X += b[i]*tl.fold(M[i],i,T.shape)
X = X/np.sum(b)
for i in range(T.shape[0]):
for j in range(T.shape[1]):
if Omega[i, j, :].all() != 0:
X[i, j, :] = T[i, j, :]
return X
def silrtc(X,b,K,O):
a = np.random.dirichlet(np.ones(3),size=1)
a = a[0,:]
M = [0, 0, 0]
for i in range(K):
X1 = tl.unfold(X, 0)
X2 = tl.unfold(X, 1)
X3 = tl.unfold(X, 2)
u1, s1, v1 = LA.svd(X1, full_matrices=False)
u2, s2, v2 = LA.svd(X2, full_matrices=False)
u3, s3, v3 = LA.svd(X3, full_matrices=False)
U = [u1,u2,u3]
V = [v1,v2,v3]
S = [s1,s2,s3]
for j in range (3):
M[j] = shrinkage_op(U[j],V[j],S[j],a[j]/b[j],X.shape)
X = eq_svtn(X,b,M,O)
return X
im = cv2.imread('target.jpg')
original_shape = im.shape
zzo = np.zeros(im.shape)
shp = (int(original_shape[0]),int(mt.floor(0.9*original_shape[1])),int(original_shape[2]))
shp2 = (int(original_shape[0]),int(mt.ceil(0.1*original_shape[1])),int(original_shape[2]))
zz = np.zeros(shp)
oo = 0xFF*np.ones(shp2)
cct = np.concatenate((zz,oo),1)
np.random.shuffle(cct.reshape(-1, 3))
cct = cct.astype(np.uint8)
im2 = np.bitwise_and(im,cct)
b = [0.0001, 0.0001,0.0001]
xx = silrtc(im2,b,64,cct)
为3个浮点数组，b为一组二维数组（矩阵），M为其中包含黑色像素的图像（3-D阵列，张量），T是删除了黑色像素的彩色图像。