print(type(Images))
ref_idx, idx, max_img = 1, 0, len(Images)
print("The image {} chosen as reference".format(ref_idx))
pano = np.asarray(Images[ref_idx])
pt_ref, descri_ref = sift_test(Images[ref_idx])
while True:
if len(Images) == idx: break
if idx == ref_idx:
idx += 1
continue
img = Images[idx]
pts, descr = sift_test(img)
M = SIFTSimpleMatcher(descr, descri_ref, Thre)
if len(M) < 3:
Images.update({max_img: img})
continue
H = RANSACFit(pts, pt_ref, M)
pano = PairStitch(Image.fromarray(np.asarray(img)),
Image.fromarray(pano),
H,
save=False,
get_array=True)
pt_ref, descri_ref = sift_test(pano)
pano = np.asarray(pano)
idx += 1
result = Image.fromarray(pano)
result.save(saveFileName)
print('The completed file has been saved as ' + saveFileName)
if (max(img.size) > 1000 or len(imgList) > 10):
img.thumbnail((np.asarray(img.size) * RESIZE).astype('int'),
Image.ANTIALIAS)
Images.update({idx: img})
print('Images loaded. Beginning feature detection...')
#%% Feature detection
Descriptor = {}
PointInImg = {}
for idx, (key, img) in enumerate(sorted(Images.items())):
I = np.asarray(img.convert('L')).astype('single')
[f, d] = sift(I, compute_descriptor=True, float_descriptors=True)
pointsInImage = swapcolumn(f[:, 0:2])
PointInImg.update({idx: pointsInImage})
Descriptor.update({idx: d})
#%% Compute Transformation
Transform = {}
for idx in range(len(imgList) - 1):
print('fitting transformation from ' + str(idx) + ' to ' + str(idx + 1) +
'\t')
M = SIFTSimpleMatcher(Descriptor[idx], Descriptor[idx + 1], Thre)
print('matching points:', len(M, ), '\n')
Transform.update({idx: RANSACFit(PointInImg[idx], PointInImg[idx + 1], M)})
#%% Make Panoramic image
print('Stitching images...')
MultipleStitch(Images, Transform, saveFileName)
print('The completed file has been saved as ' + saveFileName)
plt.imshow(Image.open(saveFileName))
- descriptors : F x 128 array
F is the number of keypoints (frames) used. The 128 length vectors
per keypoint extracted, uint8 by default. Only returned if compute_descriptors=True
Ref: https://github.com/menpo/cyvlfeat/blob/master/cyvlfeat/sift/sift.py
'''
I = np.asarray(img1.convert('L')).astype('single') # rgb2gray
[f, desc1] = sift(I, compute_descriptor=True, float_descriptors=True)
pointsInImage1 = swapcolumn(f[:, 0:2])
I = np.asarray(img2.convert('L')).astype('single') # rgb2gray
[f, desc2] = sift(I, compute_descriptor=True, float_descriptors=True)
pointsInImage2 = swapcolumn(f[:, 0:2])
#%% Matching
M = SIFTSimpleMatcher(desc1, desc2)
#%% Transformation
maxIter = 500
maxInlierErrorPixels = 0.01 * len(np.asarray(img1))
seedSetSize = np.max([3, np.ceil(0.5 * len(M))])
minInliersForAcceptance = np.ceil(0.2 * len(M))
H = RANSACFit(pointsInImage1, pointsInImage2, M, maxIter, seedSetSize,
maxInlierErrorPixels, minInliersForAcceptance)
#%% Make Panoramic image
Pano = PairStitch(img1, img2, H, saveFileName)
print('Panorama was saved as uttower_pano.jpg', saveFileName)
plt.imshow(Image.open(saveFileName))