With the explosion of digital imaging systems, high image noise levels, particularly motion blur and sensor noise, limit the applications of mobile cameras. In this paper, a novel restoration approach is developed. By uniformly splitting the exposure time and averaging the multiple under-exposed captures, a single image is obtained in which both types of noise are mitigated. Comprehensive experiments demonstrate the successful image restoration under various exposure conditions. Exposure-splitting is further optimized to fully recover the image quality under severe motion blur. This technique can be easily implemented into contemporary digital imaging systems.