Research And Implementation Of Coal Flame Image Hardware Dehaze Based On ZYNQ

The economical and stable operation of kiln was more required because of the increasing pressure on energy saving and emission reduction. Usually condition of the kiln can be quickly and directly obtained by analyzing its flame image. However, the accuracy of image feature extraction and artificial judgment based on the image processing is affected due to the complex dust inside the kiln. Therefore, it is significant and necessary to remove the dust from coal flame images.Using the dark channel prior algorithm to enhance the kiln flame image was proposed by analyzing the relevance of the kiln flame and atmospheric scattering physical model. Since the algorithm can not achieve real-time processing of high resolution image, the zynq-7000 series FPGA which contains the arm core was proposed to achieve algorithm acceleration. Around the ZYNQ series FPGA, this paper carries out the following research.(1) Using dark channel prior method for coal flame image enhancement was put forward by analyzing the common points of the image of coal flame and outdoors from the view of physical model. Dealing with the defects when using existing dark channel prior method to deal with the coal flame image, a set of empirical parameters was given to modify the value of environment light A and the transmittance. Experiments prove that the improved algorithm can get better results.(2) On this basis, the IP core of the improved algorithm was generated by studying the time complexity and the sequential relationship of each computing module. According to the ZYNQ series FPGA, the dehazing system was built by using the IP core mentioned above.(3) Finally, the dehazing system was tested and verified. The unit performance power consumption is reduced by 2 orders of magnitude compared to other methods. Moreover the processing speed is about 10 times the PC platform, ARM platform 60 times, GPU platform 5 times..