ABSTRACT
With the development of technology, computer vision systems make our life more convenient. For example, the smart car uses an image sensor to recognize pedestrians or traffic signs, but the recognition error rate of the image sensor increases in bad weather conditions, such as in fog or haze. It is important to have a clear scene image to decrease the recognition error rate and reduce the risk of accidents. In this study, we implement a dehazing system based on an atmospheric scattering model, and on a Field-Programmable Gate Array (FPGA) using a convolutional neural network. The key to achieving haze removal is to estimate a medium transmission map which indicates the light transmission rate under the medium for an input haze image. To reduce the computation time, the input image is downscaled to estimate its haze feature in order to obtain its corresponding medium transmission map. Then the estimated medium transmission map is upscaled to be the same size. The upscaled medium transmission map is used to remove the haze from the input image. To evaluate the effectiveness of our work, both of the software versions and the FPGA version results are compared. It is shown that they are consistent. Our system is implemented on an Altera DE2–115 with a camera module, and the frame rate is 5 fps under a 100 MHz clock rate.
