MindCam: An Approach for Sketch Based Image Retrieval
Sini Thankachan · International Journal of Information Systems and Computer Sciences · 2019
Image retrieval plays a major role in computer vision applications.FaceSketch Based Image Retrieval (FSBIR) draws attention with its wide application in criminal investigation by law enforcement agencies.Automatic retrieval of face images from police mug-shot databases guides investigators to locate or narrow down the search area.MindCam is an efficient image retrievalsystem based on face sketch queries.MindCam leverages Scale invariant Feature Transform (SIFT)to figure out the interest points and extract the features independent of scale.Binary hashing is used to compute binary hash codes from SIFT feature descriptors.This makes feature matching more efficient by computing the Hamming distance of the binary codes.Pool of matching candidates is identified based on Hamming distance and Fine Grained Matching is done on candidates in the pool.A user interactive approach is guaranteed by re-ranking of retrieved results.Reduced computational complexity makes MindCam suitable for realistic applications like mobile devices and wearable devices.System also provides audio description of results which makes the system user friendly.