Efficient use of local edge histogram descriptor

Abstract

The purpose of this paper is to show how the edge histogram descriptor for MPEG-7 can be efficiently utilized for image matching. Since the edge histogram descriptor recommended for the MPEG-7 standard represents only local edge distribution in an image, the matching performance for image retrieval may not be satisfactory. In this paper, to increase the matching performance, we propose to use the global and semi-local edge histograms generated directly from the local histogram bins. Then, the global, semi-global, and local histograms of two images are compared to evaluate the similarity measure. Since we exploit the absolute locations of edge in the image as well as its global composition, the proposed matching method is considered to be a more image content-based retrieval. Experimental results support this claim. Experiments on test images for MPEG-7 core experiment show that the proposed method yields better retrieval performance especially for semantic similarity.

Powerful Combination of Color Descriptor and LBP Descriptor for Image Retrieval

“…These image description methods often depend on the applications used. Some distinguishes descriptors reflecting the overall visual appearance of an image, such as color histogram [3], color moments [4], the co-occurrence matrix [5], edge histogram [6], and so on. These features are extracted from the whole of an image and don't give information for the specific region of image.…”

Efficient Similarity Search with a Pivot-Based Complete Binary Tree

“…Of these, we, in particular, utilized the four descriptors of SC, CS, EH and, HT whose distances are measured by the weighted L 1 distance [23], [24]. In terms of EH, instead of the prepared 80-dimensional vector, we exploited a 150-dimensional vector that includes elements related to global and semi-global edge histograms as described in [25]. In addition to the four descriptors, we utilized a new descriptor created by concatenating weighted feature vectors of these descriptors (mixed descriptor, MX for short), just as the descriptor by aggregating distances between the feature vectors is utilized in [22].…”

“…The object-pivot distance constraint and the pivot filtering correspond to the second process collecting objects within the range and the third process individually filtering objects in Algorithm 4, which can be evaluated as the number |S| of unbounded objects defined in Eq. (25), and as the number |W| of unfiltered objects defined in Eq. (24), respectively.…”

Multiple ocular diseases detection by graph regularized multi-label learning

“…For global features, four types of features are extracted: 64-dimensional color histogram (LAB) [11], 144-dimensional color auto-correlogram (HSV) [12], 73-dimensional edge direction histogram [13], and 128-dimensional wavelet texture [15]. For local features, 500-dimensional bags of visual words [16] are generated.…”