Practical Computation Acceleration for Large Size Single Image Filtering Based Haze Removal

The main defect of traditional filtering based image dehazing approaches is huge time consuming, especially for large size image haze removal. In the work, a serial of practical computation acceleration strategies for bilateral filtering based haze removal are proposed. Our approach strength lies in its wide practicability: better dehazing with very fast computation speed even for large size images. Since our approach requires less time consuming corresponding to less CPU load, it will be promising for some real time application, such as smart TV image enhancement.