Design And Implementation Of Automatic White Balance Algorithm Based On FPGA

The rapid development of the electronics industry has made machine vision deeply into our production and life,and the high demand of the market has promoted the rapid development of the field of image processing.Compared with traditional black and white industrial cameras,modern color industrial cameras can obtain more image information,Then,these rich information can be classified and processed subsequently.In the process of production and use,the images collected by the existing industrial cameras have color cast.Aiming at this problem,an automatic white balance system is designed.The system uses FPGA as the core processor to achieve image acquisition,storage,and preprocessing.and send.Compared to other platforms such as general-purpose processors(GPPs)and application-specific integrated circuits(ASICs),FPGAs offer greater flexibility.In order to meet the real-time,low power consumption,high performance and high flexibility of industrial cameras,FPGA is selected as the image processing accelerator to complete the imaging tasks of industrial cameras.The implementation of the automatic white balance algorithm requires a stable and reliable hardware platform.First of all,the main components of the industrial camera are selected;the hardware circuit is designed,including the image acquisition module,the main control module,the Ethernet communication module and the power supply module;and finally the preprocessing of the image is realized based on the built hardware platform.To achieve image preprocessing operation on the FPGA,the main purpose is to eliminate the dead pixels in the acquired image,improve the accuracy of dead pixel detection by reducing the detection radius and setting the dynamic threshold,reduce the number of false detections,and use the multiple sub-means around the dead pixels to replace the bad point values to protect the edges of the image,achieve the purpose of dead pixel correction,and ensure the correctness of the image data before the automatic white balance algorithm corrects.Based on the traditional grayscale world method and perfect reflection method,the automatic white balance algorithm is improved and designed,and the gain of color cast image correction is solved by using pixel mean and pixel maximum value,and finally the obtained gain is applied to the output of the color cast image.The test results show that the real-time processing speed of 230 frames/s is achieved when the image with a resolution of 1920×1200 is processed.The color cast image correction analysis,after processing,the average chromatic aberration of the picture is much smaller than the mean chromatic aberration of the original map,compared with the classical algorithms such as the grayscale world method,the average chromatic aberration obtained by the algorithm is also small enough,and good results have been achieved in subjective and objective evaluation,which can effectively improve the image quality and have a broad application scenario.By building a test platform,the stability test of industrial cameras and automatic white balance systems is completed.First,the improved automatic white balance algorithm and the traditional algorithm are used to process the images collected by the industrial camera,and the experimental verification shows that the image evaluation value processed by the improved automatic white balance algorithm is smaller than the traditional algorithm;Machine soft trigger test to verify that the camera and algorithm have certain stability.