bug fixes, added features to enable GUI

setup.py

@@ -40,7 +40,7 @@
 
 setup(
     name='leapct',
-    version='1.20', 
+    version='1.21', 
     author='Kyle Champley, Hyojin Kim', 
     author_email='champley@gmail.com, hkim@llnl.gov', 
     description='LivermorE AI Projector for Computed Tomography (LEAPCT)', 

setup_AMD.py

@@ -113,7 +113,7 @@
 
 setup(
     name='leapct',
-    version='1.20', 
+    version='1.21', 
     author='Kyle Champley, Hyojin Kim', 
     author_email='champley@gmail.com, hkim@llnl.gov', 
     description='LivermorE AI Projector for Computed Tomography (LEAPCT)', 

setup_ctype.py

@@ -40,7 +40,7 @@
 
 setup(
     name='leapct',
-    version='1.20', 
+    version='1.21', 
     author='Kyle Champley, Hyojin Kim', 
     author_email='champley@gmail.com, hkim@llnl.gov', 
     description='LivermorE AI Projector for Computed Tomography (LEAPCT)', 

src/cuda_utils.cu

@@ -798,8 +798,10 @@ float sum(const float* dev_lhs, const int3 N, int whichGPU)
     uint64 numberOfElements = uint64(N.x) * uint64(N.y) * uint64(N.z);
     //* Newest method which makes sure all cores are busy
     int num_gpu_cores = max(1024, getSPcores(whichGPU));
-    if (numberOfElements < uint64(num_gpu_cores))
-        num_gpu_cores = int(numberOfElements);
+    if (uint64(sqrt(double(numberOfElements))) < uint64(num_gpu_cores))
+        num_gpu_cores = int(sqrt(double(numberOfElements)));
+    //if (numberOfElements < uint64(num_gpu_cores))
+    //    num_gpu_cores = int(numberOfElements);
     uint64 maxNumItemsPerCore = uint64(ceil(double(numberOfElements) / double(num_gpu_cores)));
     //printf("number of cores = %d, number of chunks = %d\n", num_gpu_cores, int(numChunks));
 
@@ -859,8 +861,10 @@ float innerProduct(const float* dev_lhs, const float* dev_rhs, const int3 N, int
     uint64 numberOfElements = uint64(N.x) * uint64(N.y) * uint64(N.z);
     //* Newest method which makes sure all cores are busy
     int num_gpu_cores = max(1024, getSPcores(whichGPU));
-    if (numberOfElements < uint64(num_gpu_cores))
-        num_gpu_cores = int(numberOfElements);
+    if (uint64(sqrt(double(numberOfElements))) < uint64(num_gpu_cores))
+        num_gpu_cores = int(sqrt(double(numberOfElements)));
+    //if (numberOfElements < uint64(num_gpu_cores))
+    //    num_gpu_cores = int(numberOfElements);
     uint64 maxNumItemsPerCore = uint64(ceil(double(numberOfElements) / double(num_gpu_cores)));
     //printf("number of cores = %d, number of chunks = %d\n", num_gpu_cores, int(numChunks));
 

src/leap_preprocessing_algorithms.py

@@ -162,7 +162,7 @@ def gain_correction(leapct, g, air_scan, dark_scan, calibration_scans=None, ROI=
         #g[:,:,:] = g[:,:,:] / postageStamp[:,None,None]
         g[:] = g[:] / postageStamp[:,None,None]
 
-    badPixelCorrection(leapct, g, badPixelMap, 5, isAttenuationData=False)
+    badPixelCorrection(leapct, g, None, None, badPixelMap, 5, isAttenuationData=False)
 
     return True
 
@@ -265,7 +265,7 @@ def makeAttenuationRadiographs(leapct, g, air_scan=None, dark_scan=None, ROI=Non
 
     return True
 
-def badPixelCorrection(leapct, g, badPixelMap=None, windowSize=3, isAttenuationData=True):
+def badPixelCorrection(leapct, g, air_scan=None, dark_scan=None, badPixelMap=None, windowSize=3, isAttenuationData=True):
     r"""Removes bad pixels from CT projections
     
     LEAP CT geometry parameters must be set prior to running this function 
@@ -308,6 +308,10 @@ def badPixelCorrection(leapct, g, badPixelMap=None, windowSize=3, isAttenuationD
         #print('Estimated ' + str(np.sum(badPixelMap)) + ' bad pixels')
 
     leapct.badPixelCorrection(g, badPixelMap, windowSize)
+    if air_scan is not None:
+        leapct.badPixelCorrection(air_scan, badPixelMap, windowSize)
+    if dark_scan is not None:
+        leapct.badPixelCorrection(dark_scan, badPixelMap, windowSize)
     if isAttenuationData:
         leapct.negLog(g)
     return True
@@ -503,10 +507,13 @@ def ringRemoval_fast(leapct, g, delta=0.01, numIter=30, maxChange=0.05):
         g_sum = np.zeros((1,g.shape[1],g.shape[2]), dtype=np.float32)
         g_sum[0,:] = np.mean(g,axis=0)
     g_sum_save = leapct.copyData(g_sum)
+    minValue = np.min(g_sum_save)
+    minValue = min(minValue, 0.0)
 
     numNeighbors = leapct.get_numTVneighbors()
     leapct.set_numTVneighbors(6)
     leapct.diffuse(g_sum,delta,numIter)
+    g_sum[g_sum<minValue] = minValue
     leapct.set_numTVneighbors(numNeighbors)
 
     gainMap = g_sum - g_sum_save
@@ -675,6 +682,10 @@ def parameter_sweep(leapct, g, values, param='centerCol', iz=None, algorithmName
     if param == 'tau' and leapct.get_geometry() == 'PARALLEL':
         print('Error: tau does not apply to parallel-beam data.')
         return None
+    if leapct.get_geometry() == 'MODULAR':
+        if param == 'tau' or param == 'centerCol' or param == 'centerRow':
+            print('Error: centerCol, centerRow, and tau do not apply to modular-beam data.')
+            return None
 
     if has_torch == True and type(g) is torch.Tensor:
         f_stack = torch.zeros((len(values), leapct.get_numY(), leapct.get_numX()), dtype=torch.float32)
@@ -710,11 +721,15 @@ def parameter_sweep(leapct, g, values, param='centerCol', iz=None, algorithmName
     for n in range(len(values)):
         print(str(n) + ': ' + str(param) + ' = ' + str(values[n]))
         if param == 'centerCol':
+            #col_shift = (leapct_sweep.get_centerCol() - values[n])*leapct_sweep.get_pixelWidth()
+            #leapct_sweep.shift_detector(0.0, col_shift)
             leapct_sweep.set_centerCol(values[n])
         elif param == 'centerRow':
-            if leapct_sweep.get_geometry() == 'CONE':
+            if leapct_sweep.get_geometry() == 'CONE' or leapct_sweep.get_geometry() == 'CONE-PARALLEL':
                 z_shift = (leapct_sweep.get_centerRow() - values[n])*leapct_sweep.get_pixelHeight()*leapct_sweep.get_sod()/leapct_sweep.get_sdd()
                 leapct_sweep.set_offsetZ(leapct_sweep.get_offsetZ() + z_shift)
+            #row_shift = (leapct_sweep.get_centerRow() - values[n])*leapct_sweep.get_pixelHeight()
+            #leapct_sweep.shift_detector(row_shift, 0.0)
             leapct_sweep.set_centerRow(values[n])
         elif param == 'tau':
             delta_tau = values[n] - leapct_sweep.get_tau()