Pixel‐level snakes on the CNNUM: algorithm design, on‐chip implementation and applications

Abstract

SUMMARYIn this paper, a new algorithm for the cellular active contour technique called pixel-level snakes is proposed. The motivation is twofold: on the one hand, a higher e ciency and exibility in the contour evolution towards the boundaries of interest are pursued. On the other hand, a higher performance and suitability for its hardware implementation onto a cellular neural network (CNN) chip-set architecture are also required. Based on the analysis of previous schemes the contour evolution is improved and a new approach to manage the topological transformations is incorporated. Furthermore, new capabilities in the contour guiding are introduced by the incorporation of in ating=de ating terms based on the balloon forces for the parametric active contours. The entire algorithm has been implemented on a CNN universal machine (CNNUM) chip set architecture for which the results of the time performance measurements are also given. To illustrate the validity and e ciency of the new scheme several examples are discussed including real applications from medical imaging.

A Method Using CA-Based Pixel-Level Snakes for Multiple-Contour Expansion

“…PLS [2][3][4][5] is an evolved version of the original snake algorithm. In PLS algorithm, two forces from the internal energy and the external potential are calculated, and the motion of the curve is determined at the pixel level by the dominant force between these two forces.…”

“…The conventional PLS algorithm [2][3][4][5] is based on a CNN [8]. A CNN is a kind of neural network whose connectivity is only restricted to adjoining cells.…”

“…The active contour method (or also known as the "snake" algorithm) is one of the methods for object extraction [1]. "Pixel-level snakes" (PLS) [2][3][4][5] is a kind of snake algorithm. As for PLS, the development of the active contours is governed by a balance of two forces: one from an internal energy of the active contours and one from an external potential of a processing image.…”

Focal-plane moving object segmentation for realtime video surveillance

“…Figure 2 shows a block diagram with the main modules to be implemented in a PLS algorithm. A comprehensive review of PLS can be found in [6].…”

“…PLS have been implemented on hardware architectures with pixel-parallel processing capabilities, such as ACE4K [6,7] (based on cellular neural network paradigm) and SCAMP-3 [8,9] (based on SIMD processor array). As a consequence a high speed of processing is achieved.…”