Method and system for identifying an object in an electronically acquired image

#This invention relates a computer system for the identification of objects within electronically acquired imagery. More specifically, this invention relates to an algorithm for resolving an image comprised of a multiplicity of pixels into separate groups of pixels, wherein each pixel group corresponds to a distinct physical object contained within the image.

#The general problem of using electronic computers to analyze and interpret electronically acquired imagery involves the analysis of images that are represented by a multiplicity of individual brightness samples distributed geometrically in space. These samples are often called pixels. Each pixel represents a numerical measurement of luminosity within a small solid angle with respect to the camera. In many cases, the problem of interpreting images composed of such multiplicities of pixels is identical to the problem of associating groups or subsets of pixels within the image with physical objects within the field of view of the camera or imaging instrument which made the image.

#In many cases, the great majority of pixels within a given image bear little or no useful information. The information bearing pixels within a practical image will often be those pixels which fall along the boundaries of physical objects within the image, or along the boundaries of component parts of objects within the image.

#FIG. 1 is a reference image that illustrates gradient runs and Grenze Sets identified using one embodiment of the present invention.

#FIG. 2 is an expanded view of the highlighted area of interest of the image of FIG. 1 showing a vertically-oriented gradient run identified using one embodiment of the present invention.

#FIG. 3 is an expanded view of the highlighted area of interest of the image of FIG. 1, showing a Grenze Set defined by common boundary segments of gradient runs identified in the area of interest using an embodiment of the present invention.

#The present disclosure relates to a system and method for identifying specific regions of an electronically acquired image. The boundaries of the regions are distinguished on the basis of their brightness relative to the brightness of the neighboring pixels which are not positioned on the boundary lines. The regions are comprised of those pixels that fall on the boundary lines, and comprise groups of pixels having within which the brightness or color varies rapidly.

#Much of the information present in an image is in the form of brightness gradients that exist within the image. The concept of a “brightness gradient” cannot be understood in terms of a single pixel. A “brightness gradient” is related to the mathematical concept of “gradient.” A gradient is, by definition, a vector quantity, which is to say it has magnitude and direction. A brightness gradient has magnitude and direction as well. Because an electronically acquired image is an approximation of the actual physical scene (approximated by means of decomposition into discrete numerical samples, each of which represents the luminosity of the scene at one point), the “brightness gradients” which comprise the information-bearing parts of the electronically acquired image are also approximations by virtue of this sampling process. In this disclosure, “brightness gradients” correspond to brightness as characterized by a single composite value which is the summation of all wavelengths of light to which the electronic camera is sensitive (as in the case of “monochrome” cameras, whether they are sensitive to wavelengths of light that are visible to the human eye, or to other wavelengths, such as in infra-red, ultra-violet, or other image sensing means), or to brightness as characterized by a collection of component wavelength ranges, as in the case of electronic cameras that are configured to acquire “color,” or multi-spectral, images. The inventive method described here is applicable to monochromatic images as well as polychromatic (or “color”) images.

#The numerical value of these boundary pixels can be characterized in terms of brightness gradients. By the term “gradient” is meant a plurality of neighboring pixels arranged in a line. Frequently this linear arrangement will take the form of a straight line oriented in a direction congruent with that of the constituent rows and columns of the pixel array which forms the image. However, the present invention is not limited to gradient pixel sets that are linear and parallel to the rows and columns of the pixel array which forms the image. The present invention is useful also in cases in which the gradient pixel sets form other geometrical shapes, such as open or closed curves, polygons, etc. Furthermore, the concept of a brightness gradient, in the case of color imagery, will correspond to a set of separate co-located gradients, this set having a quantity of elements equal to the number of constituent color components detected by the image-forming instrument.