Search

We present a way to register the uterus in monocular laparoscopy in realtime using a novel two-phase approach. This differs significantly to SLAM, which is currently the leading approach for registration in MIS when scenes are approximately rigid. In the first phase we construct a 3D model of the uterus using dense SfM. This involves a method for semi-automatically masking the uterus from background structures in a set of reference frames, which we call Mask Bootstrapping from Motion (MBM). In the second phase the 3D model is registered to the live laparoscopic video using a novel wide-baseline approach that uses many texture maps to capture the real changes in appearance of the uterus. Capturing these changes means that registration can be performed reliably without needing temporal priors, which are needed in SLAM. This simplifies registration and leads to far fewer tuning parameters. We show that our approach significantly outperforms SLAM on an in vivo dataset comprising three human uteri.

Low cost CMOS cameras can have an acquisition mode called rolling shutter which sequentially exposes the scan-lines. When a single object moves with respect to the camera, this creates image distortions. Assuming 2D-3D correspondences known, previous work showed that the object pose and kinematics can be estimated from a single rolling shutter image. This was achieved using a suboptimal initialization followed by local iterative optimization. We propose a polynomial projection model for rolling shutter cameras and a constrained global optimization of its parameters. This is done by means of a semidefinite programming problem obtained from the generalized problem of moments method. Contrarily to previous work, our optimization does not require an initialization and ensures that the global minimum is achieved. This allows us to build automatically robust 2D-3D correspondences using a template to provide an initial set of correspondences. Experiments show that our method slightly improves previous work on both simulated and real data. This is due to local minima into which previous methods get trapped. We also successfully experimented building 2D-3D correspondences automatically with both simulated and real data.

We present a method to automatically select the regularization parameter in the two-term compound cost function used in image registration. Our method is called CFS (Constant Flow Sampling). It samples the regularization parameter using the constraint that the warp-induced image flow be of constant magnitude on average. Compared to other methods, CFS provably provides a global solution at a specified precision and within a finite number of steps. CFS can be embedded within any algorithm minimizing a two-term compound cost function depending on a regularization parameter. We report experimental results on the registration of several datasets of laparoscopic images.

Different solutions have been proposed for multiple objects tracking based on probabilistic algorithms. In this chapter, the authors propose the use of a single particle filter to track a variable number of objects in a complex environment. Estimator robustness and adaptability are both increased by the use of a clustering algorithm. Measurements used in the tracking process are extracted from a stereovision system, and thus, the 3D position of the tracked objects is obtained at each time step. As a proof of concept, real results are obtained in a long sequence with a mobile robot moving in a cluttered scene.

In this paper we present a technique for fully automatic, real-time cardiac SPECT (Single Photon Emitting Computed Tomography) and 2D ultrasound image registration. Our registration algorithm allows a physician to perform an ultrasound exam after a series of ECG-gated SPECT images has been acquired and see in real time the registration of both modalities. A specific ECG segmentation algorithm has been developed in order to associate each US image acquired on the fly with the appropriate SPECT volume representing a heart-beat instant.