Automatic extraction of control points for digital subtraction angiography image enhancement

Abstract

In this paper, a new automatic control point selection and matching technique for digital subtraction angiography image enhancement is proposed. The characteristic of this approach is that it uses features that are based on image moments and are invariant to symmetric blur, translation, and rotation to establish correspondences between matched regions from two X-ray images. The automatic extraction of control points is based on an edge detection approach and on local similarity detection by means of template matching according to a combined invariantsbased similarity measure. A new strategy was developed in which a 3-D space-time motion detection algorithm was used for selecting movement points (MPs) belonging to moving structures. The proposed technique has been successfully applied to register several clinical data sets including coronary applications. The experimental results demonstrate the efficiency and accuracy of the algorithm which have outperformed manual registration in terms of root mean square error at the MPs. In addition, the results of the proposed registration algorithm, using the combined invariants-based similarity measure are compared to those of the same proposed algorithm but using the energy of the histogram of differences (EHD) measure. These results clearly indicate that the combined invariants-based measure may be better suited for subpixel registration as it produces more accurate results than EHD.

Projection Operators and Moment Invariants to Image Blurring

“…These invariants, along with the centrosymmetry assumption, have been adopted by numerous researchers. They (as well as their equivalent counterparts in Fourier domain) have become very popular image descriptors and have found a number of applications, namely in matching and registration of satellite and aerial images [7], [9], [10], [11], [12], in medical imaging [13], [14], [15], in face recognition on outof-focus photographs [6], in normalizing blurred images into canonical forms [16], [17], in blurred digit and character recognition [18], in robot control [19], in image forgery detection [20], [21], in traffic sign recognition [22], [23], in fish shape-based classification [24], in wood industry [25], [26], in weed recognition [27], in cell recognition [28] and in focus/defocus quantitative measurement [29]. In the last few years yet another broad application area of blur invariants has appeared.…”

“…Since their first appearance in 1996, invariants to a centrosymmetric blur (N = 2) have been several times successfully used as the features in landmark-based registration of remotely sensed [7], [10], [33], [11], [12], [86], medical [13], [15], [14], indoor [87] and outdoor [49], [57], [55] scenes. As we demonstrate, introducing new invariants to N -FRS blur with higher discrimination power broadens their registration capability.…”

Automatic Image Registration Using Mexican Hat Wavelet, Invariant Moment, and Radon Transform

“…Moment-based invariants are the most common region-based image invariants which have been used as pattern features in many applications [27].…”

Image Deconvolution in the Moment Domain

“…Moment invariants to convolution have found numerous applications, namely in image matching and registration of satellite and aerial images [9,24,5,20,15], in medical imaging [4,3,32,2], in face recognition on out-of-focus photographs [10], in normalizing blurred images into canonical forms [34,36], in blurred digit and character recognition [21], in robot control [29], in image forgery detection [22,23], in trac sign recognition [19,18], in sh shape-based classication [35], in weed recognition [26], and in cell recognition [25]. Their popularity follows from the fact that the convolution model of image formation g(x, y) = (f * h)(x, y), (5.1) where g(x, y) is the acquired blurred image of a scene f (x, y) and the kernel h(x, y) stands for the point-spread function (PSF) of the imaging system, is widely accepted and frequently used compromise between universality and simplicity.…”