Limitations of autocorrelogram technique in CBIR system
Anucha Tungkasthan · 2017
This paper presents a popular color feature extraction's technique for image retrieval, namely the Autocorrelogram (AC). The main objective of this paper attempts to present weaknesses of the Autocorrelogram in indexing efficiency and computational complexity. In the first issue, the paper presents an existing problem and introduces two new problems that effect to the accuracy of indexing and retrieval images. In the another issue, this paper presents the complexity of existing similarity measurement scheme, which directly affects to the computation time for searching images in a database, especially in large image databases. In addition, the paper suggests the technique on how to improve the performance of Autocorrelogram. It can be used as a guideline for effective implementation when it is used in real world application.