Python scipy.ndimage 模块，affine_transform() 实例源码

def Run(self, img_path, guide_image_path='', objective=0):
        """Run deep dream"""
        self.guide_path = guide_image_path
        if self.guide_path != '':
            self.Get_guide()
        self.net.blobs.keys()
        if img_path != '':
            frame = PIL.Image.open(img_path)
            frame = imresize(frame)
            frame = np.float32(frame)
        else:
            frame = self.GenerateInputImage()
        frame_i = 0
        h, w = frame.shape[:2]
        #s = 0.05 # scale coefficient
        for i in xrange(self.epoch):
            start = time.time()
            frame = self.Deepdream(frame)
            PIL.Image.fromarray(np.uint8(frame)).save("frames/%04d.jpg"%frame_i)
            #frame = nd.affine_transform(frame, [1-s,1-s,1], [h*s/2,w*s/2,0], order=1)
            frame_i += 1
            stop = time.time()
            print "Time cost for {:d}th image: {:.3f} s".format(i,stop-start)

def Run(self, img_path, guide_image_path='', objective=0):
        """Run deep dream"""
        self.guide_path = guide_image_path
        if self.guide_path != '':
            self.Get_guide()
        self.net.blobs.keys()
        if img_path != '':
            frame = PIL.Image.open(img_path)
            frame = imresize(frame)
            frame = np.float32(frame)
        else:
            frame = self.GenerateInputImage()
        frame_i = 0
        h, w = frame.shape[:2]
        #s = 0.05 # scale coefficient
        for i in xrange(self.epoch):
            start = time.time()
            frame = self.Deepdream(frame)
            PIL.Image.fromarray(np.uint8(frame)).save("frames/%04d.jpg"%frame_i)
            #frame = nd.affine_transform(frame, [1-s,1-s,1], [h*s/2,w*s/2,0], order=1)
            frame_i += 1
            stop = time.time()
            print "Time cost for {:d}th image: {:.3f} s".format(i,stop-start)

def image_in_image(im1,im2,tp):
    m, n = im1.shape[:2]
    fp = np.array([[0,m,m,0],[0,0,n,n],[1,1,1,1]])

    # compute affine transform and apply
    H = homography.Haffine_from_points(tp,fp)
    im1_t = ndimage.affine_transform(im1,H[:2,:2],
                    (H[0,2],H[1,2]),im2.shape[:2])
    alpha = (im1_t > 0)

    return (1-alpha)*im2 + alpha*im1_t

def run(self, ips, snap, img, para = None):
        if para == None: para = self.para
        a = para['ang']/180.0*np.pi
        o = np.array(ips.size)*0.5
        if ips.roi!=None:
            box = ips.roi.get_box()
            o = np.array([box[1]+box[3],box[0]+box[2]])*0.5
        trans = np.array([[np.cos(a),-np.sin(a)],[np.sin(a),np.cos(a)]])
        offset = o-trans.dot(o)
        nimg.affine_transform(snap, trans, output=img, offset=offset)

def run(self, ips, snap, img, para = None):
        if para == None: para = self.para
        k = 1/para['zoom']
        o = np.array(ips.size)*0.5
        if ips.roi!=None:
            box = ips.roi.get_box()
            o = np.array([box[1]+box[3],box[0]+box[2]])*0.5
        trans = np.array([[k,0],[0,k]])
        offset = o-trans.dot(o)
        nimg.affine_transform(snap, trans, output=img, offset=offset)

def run(self, ips, img, buf, para = None):
        if para == None: para = self.para
        a = para['ang']/180.0*np.pi
        trans = np.array([[np.cos(a),-np.sin(a)],[np.sin(a),np.cos(a)]])
        o = np.array([self.para['oy'], self.para['ox']])
        offset = o-trans.dot(o)
        if self.para['img']:
            nimg.affine_transform(img, trans, output=buf, offset=offset)
        if self.para['msk'] and self.bufroi!=None:
            ips.roi = self.bufroi.affine(trans, o[::-1]-trans.dot(o[::-1]))

def run(self, ips, img, buf, para = None):
        if para == None: para = self.para
        self.count(False)
        trans = np.array([[1/self.para['ky'],0],[0,1/self.para['kx']]])
        o = np.array([self.para['oy'], self.para['ox']])
        offset = self.orio[::-1]-trans.dot(o)
        if self.para['img']:
            nimg.affine_transform(img, trans, output=buf, offset=offset)
        trans = np.array([[self.para['kx'],0],[0, self.para['ky']]])
        offset = o[::-1]-trans.dot(self.orio)
        if self.para['msk'] and self.bufroi!=None:ips.roi = self.bufroi.affine(trans, offset)
        if self.para['img'] and not ips.get_msk('out') is None: 
            buf[ips.get_msk('out')] = img[ips.get_msk('out')]
        ips.update = True

def random_affine(transpose_scale, affine_scale, fill=0):
    def transform(glyph):
        # random transpose: +-transpose_max
        transpose = np.random.normal(loc=0, scale=transpose_scale, size=2)

        # identity matrix (no transform)
        affine = np.asarray([[1, 0], [0, 1]], dtype='float32')
        # add randomness: +-affine_max
        affine += np.random.normal(loc=0, scale=affine_scale, size=(2,2))

        return ndimage.affine_transform(glyph, matrix=affine, offset=transpose, cval=fill)

    return transform