Performance analysis of vehicle detection based on spatial saliency and local image features in H.265 (HEVC) 4K video for developing a relationship between iou and subjective evaluation value

Abstract

A vehicle detection method fusing spatial saliency and local image features is presented in this paper to analyze the detection performance in compressed domain. Most of the existing visual saliency models use the input images, in which salient objects are to be detected and are free from complex background and overlapping areas. Moreover, they are very sensitive to the complex scene and different illuminations. In this method, the Scale Invariant Feature Transform and Harris features in combination with spatial saliency model play an important role in detecting vehicles from the scene. A one‐to‐one symmetric search is performed on the descriptors to select a set of matched interest point pairs for vehicle detection. We use 4K video of a road scene with different types of vehicles. The proposed method is able to detect desired overlapping objects from the road scene without heavy computation like other training‐based methods. In addition, the detection performance is analyzed in a H.265 compressed domain to help construct a relationship between objective and subjective evaluation value. The performance of this method is demonstrated by the experimental results. © 2020 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.