Cellular Automata for Elementary Image Enhancement
Graphical Models and Image Processing
any previous information that can be determined by straightforward algorithms. On the other hand, some fea-We study various cellular automata as algorithms for elementary image enhancement, which refers to methods used to imtures of the image, such edges, boundaries, specific shapes prove features of an image without previous information about (lines for instance), the skeleton, the connexity property, them that can be implemented by straightforward techniques. and so on, can be used in further
... analysis of it. This is Cellular automata appear as natural tools for image processing the area of the segmentation, description, and recognition due to their local nature and simple parallel computer impletechniques in image processing. mentation. For this reason various cellular automata algorithms In this work we study some cellular automata for elemenfor sharpening and smoothing are presented and studied in tary image enhancement. It is supposed that there is no this context. Their dynamical behavior is characterized for seinformation about the features of the image that must be quential and parallel updating by associating to them strictly improved. In this context enhancement techniques appear decreasing functionals with the dynamics of the automata. as a natural requirement for noisy or blurred images as a Since tight bounds for the transient time usually cannot be obtained from these operators, a numerical study is needed to first level technique to improve them. Within this area analyze their typical performance and effects. For this purpose we found sharpening and smoothing techniques, whose we compare them with the classical methods for real two dimenpurpose is to locally increase or decrease the gray level sional images in terms of convergence rate, effects, and stability differences of the image and which can also be used for in front of noise. The cellular automata methods studied present noise-reduction without loss of the main characteristics of very fast convergence to fixed points, noise stability, and imthe image. provements on real images, which are features that allow us In Section 2, we define our cellular automata used for to propose them as a first level elementary image enhancesharpening and smoothing. We characterize their dynamiment.