A New Approach of Gray Images Binarization for Artificial Vision Systems with Threshold Methods

This paper presents some aspects of the (gray level) image binarization methods used in artificial vision systems. It is introduced a new approach of gray level image binarization for artificial vision systems dedicated to the specific class of applications for moving scene in industrial automation – temporal thresholding. In the first part of the paper are remarked some limitations of using the global optimum thresholding in gray level image binarization. In the second part of this paper are presented some aspects of the dynamic optimum thresholding method for gray level image binarization. In the third section are introduced the concepts of temporal histogram and temporal thresholding, starting from classic methods of global and dynamic optimal thresholding of the gray level images. In the final part are presented some practical aspects of the temporal thresholding method in artificial vision applications for the moving scene in robotic automation class; highlighting the influence of the acquisition frequency on the methods results. 1 IMAGE BINARIZATION WITH GLOBAL THRESHOLD Threshold methods are defined as starting from the analyse of the values of a function T of the type: T = T [x, y, p(x, y), f(x, y)] (1) Where: f(x, y) – represents the intensity value of the image element located on the co-ordinates (x, y); p(x,y) – represents the local properties of the specific point (like the average intensity of a region centred in the co-ordinates (x, y)). T – is the binarization threshold The goal is to obtain from an original gray level image, a binary image g(x, y) defined by: ⎩⎨ ⎧ ≤ > = T y x f T y x f y x g ) , ( for 0 ) , ( for 1 ) , ( (2) For T a function only of f(x, y), the obtained threshold is called global threshold. In the case of T a function of both f(x, y) and p(x, y), the obtained threshold is named local threshold. In the case of T a function of all f(x, y), p(x, y), x and y, the threshold is a dynamic threshold. 1.1 Intensity Level on Normal Distribution Assumption Gray level histogram represents the probability density function of the intensity values of the image. In order to simplify the explanations, we suppose the image histogram of the gray levels is composed from two values combined with additive Gaussian noise: The first segment of the image histogram corresponds to the background points – the intensity levels are closer to the lower limit of the range (the background is dark) The second segment of the image histogram corresponds to the object points – the intensity levels are closer to the upper limit of the intensity range (the objects are bright). The problem is to estimate a value of the threshold T for which the image elements with an intensity value lower than T will contain background points and the pixels with the intensity value greater than T will contain object points, with a minimum error. For a real image, the partitioning between the