Cellular Neural Networks (CNNs) constitute a class of nonlinear, recurrent and locally interconnected arrays. These arrays consist of identical dynamical cells (neurons] that operate in parallel. The connection between cells (synapse) is controlled by set of parameters. These parameters set, are called templates, determines exclusively CNN dynamic behavior. The operation of the CNNs depends primarily on these templates coefficients. Image processing via CNNs only becomes efficient if the network is implemented in analog circuits. The VLSI implementation of the CNN analog circuit is designed based on Multiple-Input Operational Transconductance Amplifier (MiOTA), using CMOS technology. MiOTA circuit generates an output current which is a linear sum of a number of weighted inputs. The weight of each input (synapse weight) is controlled by ,a bias voltage, in addition to the aspect ratio of the transistors of each synapse. That gives the ability of Programming the CNN circuit by different templates. In this work; the CNN cell structure, dynamic range, and the stability analysis are discussed. Also a practical survey of the design rules of uncoupled and coupled CNN templates with binary inputs and outputs is given. CNN templates are designed and tested for several image processing tasks, using MATLAB programs. The problem of VLSI implementation of CNNs based on CMOS MiOTA is formulated and investigated. Also the analog CNNs circuits that designed for selective image processing operations are simulated, using OrCAD PSPICE program.