Abstarct: Today, image processing methods have received much attention from researchers and are used to improve the obtained images. One of the problems that may arise in outdoor images is fog caused by water and dust particles in the air. This phenomenon causes the radiated light to scatter and the resulting image to be cloudy and opaque. Since in machine vision methods, clear images without distortion are needed, it is very necessary to remove fog from images using image processing methods. Many achievements have been obtained in the haze elimination in the recent years. The dark channel is one of these methods which also used in this thesis. The first step of this technique is creation of dark channel of the corresponding hazing image. Then, the atmospheric light and the image transmission map, which shows the haze in the image, would be obtained. Finally, by use of these values and also applying the inverse of the physical model ,the image could be dehazed. Since use of the dark channel method leads to have the dehazed image darker than the main one, the adaptive gamma correction is considered in this thesis to deal with the mentioned issue. Likewise, it is attempted to apply the proposed method on FPGA to benefit the advantages of parallelizing and the high speed of the FPGA chip. The suggested approach is applied to the Vertix6 chip and images with the size of 2456*1384 pixel and the rate of 60 frxame/second are dehazed. The maximum frequency of this method which makes it superior in comparison with the previous studied methods is 216 MHz. In additionally, the achieved results from the quality standards of MSE, PSNR, and SSIM, which are 744, 19.49 and 0.8370 ,respectively, in the new proposed method confirm the dehazed images improved in comparison with the previous studies.