COMPUTERIZED VISION TECHNIQUES Artificial vision is the science that develops the theoretical and algorithmic foundations through which useful information about the world can be automatically extracted and analyzed by a observed image, a set of images, or sequence of images from calculations made by computers for special or general purposes (notes). The main goal of applying computer vision is to produce automated recognition systems that can be equivalent to or possibly better than human performance. Computer vision can be used to enable new relationship techniques and connect both the physical and virtual worlds. There are some computer vision techniques that are used to produce the best quality computer vision application. Image Processing Image processing is a technique in which an image data is digitized and various mathematical operations are applied to the data, usually with a digital computer, in order to create an improved image that is more useful or pleasing for a human observer, or to perform some of the interpretation and recognition tasks usually performed by humans. Image processing is separated into two levels: lower level processing and higher level processing. Lower-level processing takes image pixels as input and performs tasks such as image enhancement, feature extraction, and image segmentation. The higher level processing takes the output of the lower level processing as input and generates output related to the system. An example of tasks performed in higher level processing includes vehicle tracking (T., 1998). In general, image processing operations can be classified into four types (Y., 2010), namely:1. Pixel operations: The output on a pixel depends only on the input on that pixel, regardless of the center of the paper, and computes the median M of the squared differences between the corresponding points and the transformed points. Then select affine parameters for which the median of the squared difference is the minimum. According to the above procedure, the probability p of obtaining at least one background data set and corresponding points is obtained from the following equation.〖p(ε ,q,M)=1-(1-((1-ε) q)〗^3) ͫWhere ε(<0.5) is the ratio of the regions of moving objects to the entire image and q is the probability that the corresponding points are positioned correctly Find. This method will provide an accurate and reliable model. REFERENCES Behard, A., Shahrokni, A., & Motanedi, S, A. (2002). A robust vision-based moving target detection and tracking system. Kaukorata, T., & Smed, J. (2006). Role of the pattern recognition platform.T., LC (1998). Image processing and pattern recognition. Elsevier.