Research On Application Of Image Processing Based On Cellular Neural Networks In Imaging Measurement Of Displacement

Since the cellular neural networks were proposed, they have achieved greatdevelopment. Due to their high-speed, real-time parallel computing and easyimplementation of VLSI, cellular neural networks have been widely used in imageprocessing, pattern recognition, associative memory, secure communication and otherareas. The displacement measurement system based on imaging is an effective methodfor deformation measurement of large-scale infrastructure such as bridges. Rapid andaccurate recognition of the cursor image are the key of the measurement system incomplex natural environment. Conventional image processing methods which areadopted as the main solution in current measurement system are complicated andcomputationally expensive. And it’s very difficult for them to meet the requirements ofrecognizing precision and processing speed in real-time monitoring system. Due to theadvantages of cellular neural networks in image processing, the image processingtheories and methods based on cellular neural networks are applied to displacementmeasurement system. The application of cellular neural networks has improved therecognizing precision and processing speed of cursor image in complex naturalenvironment. It has great value on displacement monitoring of large-scale infrastructuresuch as bridge.The main contents of this paper are listed as follows:①The concept and research status of cellular neural networks are introduced. Thedynamic range and stability are analyzed in details according to its structuralcharacteristics and mathematical models, and then the basic idea of cellular neuralnetworks applied to image processing is also introduced.②A novel design method of cellular neural networks templates is proposed based onimproved Particle Swarm Optimization algorithm. The method realizes the design of thenoise removal template and the edge detection template which have a better imageprocessing effects than traditional methods. And the obtained templates are applied tothe processing of laser spot image in bridge pylon displacement measurement system.③An image processing solution based on cellular neural network is designed for thedisplacement measurement system of bridge pylon. It contains an fast search algorithmfor the threshold is proposed and applied to image thresholding based on cellular neuralnetworks. The morphological templates based on cellular neural networks are also designed in the solution, and the designed templates are suitable for the structure ofarbitrary shape element. With the use of circle fitting algorithm to locate the center oflaser spot，the positioning accuracy of the center is improved.④For the vertical displacement measurement system based on bar lampself-calibration imaging, a novel thinning algorithm of binary image based on cellularneural networks is propose. The novel thinning algorithm can thin the binary lampimage effectively and maintains the connectivity. Then a solution of verticaldisplacement measurement is designed based on combining the thinning algorithm withthe thresholding and morphological algorithm. Compared to traditional processingmethods, the proposed solution improves the measurement accuracy of the system.Finally, the conclusion is summed up. Some deficiencies as well as further work arealso given.