Skip to Main Content
This paper describes an anomaly target detection algorithm based on JPEG images. The algorithm deals directly with JPEG image data eliminating JPEG image decoding process. The algorithm extracts the JPEG DC parameters, and then compares two frames of adjacent DC parameters to determine whether the abnormal targets appear. Compared to the YUV image data processing methods, it costs less storage space, access time and computation, which is more suitable for embedded systems with limited resources. Finally, the simulation results of the algorithm are given.