Analysis of image similarity with CBIR concept using wavelet transform and threshold algorithm
Abdul Haris Rangkuti, Nashrul Hakiem, Rizal Broer Bahaweres, Agus Harjoko, Agfianto Eko Putro · 2013
This paper carried out to develop the concept of CBIR by using Wavelet Transform Methods for feature form and texture extraction and adaptive histogram for feature color extraction. The method is not only recognize the images that have been stored in database, but also be able to find some resemblance ornament image or texture as well as form. Although They are different size, direction of slope, and the layout of texture and color. In calculating the percentage of similarity is not only based on performance measurement precision but also the image of the relevant. Basically Grade value calculation is using the fuzzyfication process to get the similarity values with the S-curve which is then used as input to perform retrieval of image with the threshold algorithm. It will display the image based on the representation of the highest grade in each query image, which has been compared with the image database. High grade values indicates that the characteristic image of the sample (query) is similar to the image database and others. After that proceed by comparing the value of grade representation of the image by using the min operator in fuzzy logic. The Advantage threshold algorithm is the simplicity of similarity image process when the performance of CBIR becomes more reliable.