Design And Implementation Of An Adaptive Median Filter In Intelligent Monitoring Systems

Intelligent monitoring system needs to make good on the image filter processing has been collected before positioning,tracking and understanding to a specific target. Because in the industrial, agricultural, commercial and public service places in different complex environment field, image sensor, the decoding processing device and transmission channels can produce noise, and the communication system because of its defects and electromagnetic and other factors will also produce a lot of pulse jamming. Image noise filtering design of median filter is introduced in the intelligent monitoring system can not only better have been collected, and to have outstanding performance in ensuring the original image and real-time image processing.In the traditional median filtering algorithm has been greatly improved, bubble algorithm original too tedious now already did not see more, learn fast median filtering algorithm in the design of the filter not only improves the computing speed, but also reduce the resource usage. Has very important value to identify the image of the edge of the image and the subsequent processing, while the ordinary median filtering in the maintenance of the original image edge details, especially in the face of various picture details still has a lot of room for improvement. Adaptive median filter algorithm reference decision points in the design, mainly for the sliding window center sampling pixel values of the maximum or minimum value of the median filter, this exploration is reduced to a certain extent. All uses neighborhood median instead of sampling malpractice pixels, to protect the image edge plays a positive role.Design of adaptive median filter in intelligent monitoring system is mainly based on the hardware platform of FPGA Altera Cyclone EP1C6Q240C8set up, combined with SAA7111A and SAA7121chip and memory, application of Altera company’s Quartus II integrated development software and Modelsim simulation software design implementation. In the design, the data read between the filtering module into a pre unit registers and BRAM to optimize the design, to ensure that the low amount of hardware resources usage. Application of dual port BRAM cache to resolve the input and output frequency mismatch problem, each from the BRAM read out a data window, data ready, register data comparison of pre unit immediately ready. In the system simulation diagram analysis and filtering of images before and after the comparison, it has achieved the requirements of design.