Enhancement of Drone Facial Images with Low Resolution Based on Super-Resolution Generative Adversarial Network (SRGAN)

Drones or Unmanned Aerial Vehicles (UAVs) are frequently employed to visit remote locations or places that are inhumanely unreachable. Drones are seen as appropriate for monitoring busy or disaster-hit regions. The UAVs commonly known as Drones are used to capture images from significant distances at different angles to avoid collisions. So, these images will suffer from quality like low resolution, and blurriness due to some environmental conditions. Traditional approaches for image quality enhancement are Histogram equalization, Bipolar interpretation, De-noising, and sharpening which come under image processing and deal with the manipulation and analysis of images including image enhancement, restoration, and analysis. These methods often struggle from over-enhancement or loss of detail in an image, and sometimes may not produce visually appealing results for all images. Over 20 epochs, we assessed the performance of our model and obtained the following results, the values for the quality metrics Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), and Structural Similarity Index (SSIM) were found to be 0.63 units, 2.50 dB, and 0.43 scores, respectively.