From cdef6a981f1772ed04fe44bbe2b8251983a4ba7a Mon Sep 17 00:00:00 2001 From: dkazanc Date: Wed, 27 Nov 2019 18:38:59 +0000 Subject: modifications in pdtv --- demos/demo_cpu_regularisers.py | 6 ++++-- src/Core/regularisers_CPU/PD_TV_core.c | 6 +++--- src/Core/regularisers_CPU/PD_TV_core.h | 2 +- src/Python/ccpi/filters/regularisers.py | 8 +++++--- src/Python/src/cpu_regularisers.pyx | 18 +++++++++++------- 5 files changed, 24 insertions(+), 16 deletions(-) diff --git a/demos/demo_cpu_regularisers.py b/demos/demo_cpu_regularisers.py index 9888743..09781d5 100644 --- a/demos/demo_cpu_regularisers.py +++ b/demos/demo_cpu_regularisers.py @@ -179,7 +179,8 @@ pars = {'algorithm' : PD_TV, \ 'tolerance_constant':1e-06,\ 'methodTV': 0 ,\ 'nonneg': 1, - 'lipschitz_const' : 6} + 'lipschitz_const' : 8, + 'tau' : 0.0025} print ("#############PD TV CPU####################") start_time = timeit.default_timer() @@ -189,7 +190,8 @@ start_time = timeit.default_timer() pars['tolerance_constant'], pars['methodTV'], pars['nonneg'], - pars['lipschitz_const'], 'cpu') + pars['lipschitz_const'], + pars['tau'],'cpu') Qtools = QualityTools(Im, pd_cpu) pars['rmse'] = Qtools.rmse() diff --git a/src/Core/regularisers_CPU/PD_TV_core.c b/src/Core/regularisers_CPU/PD_TV_core.c index cdce71b..65b8711 100644 --- a/src/Core/regularisers_CPU/PD_TV_core.c +++ b/src/Core/regularisers_CPU/PD_TV_core.c @@ -29,6 +29,7 @@ * 5. lipschitz_const: convergence related parameter * 6. TV-type: methodTV - 'iso' (0) or 'l1' (1) * 7. nonneg: 'nonnegativity (0 is OFF by default, 1 is ON) + * 8. tau: time marching parameter * Output: * [1] TV - Filtered/regularized image/volume @@ -37,17 +38,16 @@ * [1] Antonin Chambolle, Thomas Pock. "A First-Order Primal-Dual Algorithm for Convex Problems with Applications to Imaging", 2010 */ -float PDTV_CPU_main(float *Input, float *U, float *infovector, float lambdaPar, int iterationsNumb, float epsil, float lipschitz_const, int methodTV, int nonneg, int dimX, int dimY, int dimZ) +float PDTV_CPU_main(float *Input, float *U, float *infovector, float lambdaPar, int iterationsNumb, float epsil, float lipschitz_const, int methodTV, int nonneg, float tau, int dimX, int dimY, int dimZ) { int ll; long j, DimTotal; - float re, re1, tau, sigma, theta, lt; + float re, re1, sigma, theta, lt; re = 0.0f; re1 = 0.0f; int count = 0; //tau = 1.0/powf(lipschitz_const,0.5); //sigma = 1.0/powf(lipschitz_const,0.5); - tau = 0.02; sigma = 1.0/(lipschitz_const*tau); theta = 1.0f; lt = tau/lambdaPar; diff --git a/src/Core/regularisers_CPU/PD_TV_core.h b/src/Core/regularisers_CPU/PD_TV_core.h index b4e8a75..97edc05 100644 --- a/src/Core/regularisers_CPU/PD_TV_core.h +++ b/src/Core/regularisers_CPU/PD_TV_core.h @@ -47,7 +47,7 @@ limitations under the License. #ifdef __cplusplus extern "C" { #endif -CCPI_EXPORT float PDTV_CPU_main(float *Input, float *U, float *infovector, float lambdaPar, int iterationsNumb, float epsil, float lipschitz_const, int methodTV, int nonneg, int dimX, int dimY, int dimZ); +float PDTV_CPU_main(float *Input, float *U, float *infovector, float lambdaPar, int iterationsNumb, float epsil, float lipschitz_const, int methodTV, int nonneg, float tau, int dimX, int dimY, int dimZ); CCPI_EXPORT float DualP2D(float *U, float *P1, float *P2, long dimX, long dimY, float sigma); CCPI_EXPORT float DivProj2D(float *U, float *Input, float *P1, float *P2, long dimX, long dimY, float lt, float tau); diff --git a/src/Python/ccpi/filters/regularisers.py b/src/Python/ccpi/filters/regularisers.py index d65c0b9..bc745fe 100644 --- a/src/Python/ccpi/filters/regularisers.py +++ b/src/Python/ccpi/filters/regularisers.py @@ -53,7 +53,7 @@ def FGP_TV(inputData, regularisation_parameter,iterations, .format(device)) def PD_TV(inputData, regularisation_parameter, iterations, - tolerance_param, methodTV, nonneg, lipschitz_const, device='cpu'): + tolerance_param, methodTV, nonneg, lipschitz_const, tau, device='cpu'): if device == 'cpu': return TV_PD_CPU(inputData, regularisation_parameter, @@ -61,7 +61,8 @@ def PD_TV(inputData, regularisation_parameter, iterations, tolerance_param, methodTV, nonneg, - lipschitz_const) + lipschitz_const, + tau) elif device == 'gpu' and gpu_enabled: return TV_PD_CPU(inputData, regularisation_parameter, @@ -69,7 +70,8 @@ def PD_TV(inputData, regularisation_parameter, iterations, tolerance_param, methodTV, nonneg, - lipschitz_const) + lipschitz_const, + tau) else: if not gpu_enabled and device == 'gpu': raise ValueError ('GPU is not available') diff --git a/src/Python/src/cpu_regularisers.pyx b/src/Python/src/cpu_regularisers.pyx index 08e247c..8de6aea 100644 --- a/src/Python/src/cpu_regularisers.pyx +++ b/src/Python/src/cpu_regularisers.pyx @@ -20,7 +20,7 @@ cimport numpy as np cdef extern float TV_ROF_CPU_main(float *Input, float *Output, float *infovector, float *lambdaPar, int lambda_is_arr, int iterationsNumb, float tau, float epsil, int dimX, int dimY, int dimZ); cdef extern float TV_FGP_CPU_main(float *Input, float *Output, float *infovector, float lambdaPar, int iterationsNumb, float epsil, int methodTV, int nonneg, int dimX, int dimY, int dimZ); -cdef extern float PDTV_CPU_main(float *Input, float *U, float *infovector, float lambdaPar, int iterationsNumb, float epsil, float lipschitz_const, int methodTV, int nonneg, int dimX, int dimY, int dimZ); +cdef extern float PDTV_CPU_main(float *Input, float *U, float *infovector, float lambdaPar, int iterationsNumb, float epsil, float lipschitz_const, int methodTV, int nonneg, float tau, int dimX, int dimY, int dimZ); cdef extern float SB_TV_CPU_main(float *Input, float *Output, float *infovector, float mu, int iter, float epsil, int methodTV, int dimX, int dimY, int dimZ); cdef extern float LLT_ROF_CPU_main(float *Input, float *Output, float *infovector, float lambdaROF, float lambdaLLT, int iterationsNumb, float tau, float epsil, int dimX, int dimY, int dimZ); cdef extern float TGV_main(float *Input, float *Output, float *infovector, float lambdaPar, float alpha1, float alpha0, int iterationsNumb, float L2, float epsil, int dimX, int dimY, int dimZ); @@ -159,11 +159,11 @@ def TV_FGP_3D(np.ndarray[np.float32_t, ndim=3, mode="c"] inputData, #****************************************************************# #****************** Total-variation Primal-dual *****************# #****************************************************************# -def TV_PD_CPU(inputData, regularisation_parameter, iterationsNumb, tolerance_param, methodTV, nonneg, lipschitz_const): +def TV_PD_CPU(inputData, regularisation_parameter, iterationsNumb, tolerance_param, methodTV, nonneg, lipschitz_const, tau): if inputData.ndim == 2: - return TV_PD_2D(inputData, regularisation_parameter, iterationsNumb, tolerance_param, methodTV, nonneg, lipschitz_const) + return TV_PD_2D(inputData, regularisation_parameter, iterationsNumb, tolerance_param, methodTV, nonneg, lipschitz_const, tau) elif inputData.ndim == 3: - return TV_PD_3D(inputData, regularisation_parameter, iterationsNumb, tolerance_param, methodTV, nonneg, lipschitz_const) + return TV_PD_3D(inputData, regularisation_parameter, iterationsNumb, tolerance_param, methodTV, nonneg, lipschitz_const, tau) def TV_PD_2D(np.ndarray[np.float32_t, ndim=2, mode="c"] inputData, float regularisation_parameter, @@ -171,7 +171,8 @@ def TV_PD_2D(np.ndarray[np.float32_t, ndim=2, mode="c"] inputData, float tolerance_param, int methodTV, int nonneg, - float lipschitz_const): + float lipschitz_const, + float tau): cdef long dims[2] dims[0] = inputData.shape[0] @@ -190,6 +191,7 @@ def TV_PD_2D(np.ndarray[np.float32_t, ndim=2, mode="c"] inputData, lipschitz_const, methodTV, nonneg, + tau, dims[1],dims[0], 1) return (outputData,infovec) @@ -198,8 +200,9 @@ def TV_PD_3D(np.ndarray[np.float32_t, ndim=3, mode="c"] inputData, int iterationsNumb, float tolerance_param, int methodTV, - int nonneg, - float lipschitz_const): + int nonneg, + float lipschitz_const, + float tau): cdef long dims[3] dims[0] = inputData.shape[0] @@ -218,6 +221,7 @@ def TV_PD_3D(np.ndarray[np.float32_t, ndim=3, mode="c"] inputData, lipschitz_const, methodTV, nonneg, + tau, dims[2], dims[1], dims[0]) return (outputData,infovec) -- cgit v1.2.1