Hierarchical String Cuts: A Translation, Rotation, Scale, and Mirror Invariant Descriptor for Fast Shape Retrieval

Abstract

This paper presents a novel approach for both fast and accurately retrieving similar shapes. A hierarchical string cuts (HSC) method is proposed to partition a shape into multiple level curve segments of different lengths from a point moving around the contour to describe the shape gradually and completely from the global information to the finest details. At each hierarchical level, the curve segments are cut by strings to extract features that characterize the geometric and distribution properties in that particular level of details. The translation, rotation, scale and mirror invariant HSC descriptor enables a fast metric based matching to achieve the desired high accuracy. Encouraging experimental results on four databases demonstrated that the proposed method can consistently achieve higher (or similar) retrieval accuracies than the state-of-the-art benchmarks with a more than 120 times faster speed. This may suggest a new way of developing shape retrieval techniques in which a high accuracy can be achieved by a fast metric matching algorithm without using the time-consuming correspondence optimisation strategy.

From species to cultivar: Soybean cultivar recognition using joint leaf image patterns by multiscale sliding chord matching

“…In this research, we built a soybean For some soybean cultivars, their leaves from different parts of the plants exhibit a quite diverse appearance in completely different shapes, while for other soybean cultivars, their leaves from different parts of the plants have the same shape. However, compared to the species leaf image databases Leaf100 [9], MEW2012 [30], ICL [16], the leaves in the SoyCultivar database are highly similar due to the fact that they all belong to the same species, making it a new and challenging dataset for the pattern recognition research community. Texture Context (SPTC) [6], (5) Bag of Contour Fragments (BCF) [50], and (6) Structure…”

“…It is very encouraging to observe that the proposed method achieved an exciting accuracy of 67.50% on such a challenging cultivar leaf recognition task. It is 33.00%, 35.50%, 35.25%, 34.25%, 27.25%, 33.25%, 38.50%, and 27.50% higher than the benchmark approaches HSC [9], MDM-CD-RM [16], MDM-ID-RA [16], IDSC [6], SPTC [6], BCF+SVM [50], BCF [50]+1NN, and SIT [48] respectively. As explained in the design of the MSCM method, the proposed sliding chord measures characterize the leaf image not only from its shape, but also from its appearance which results in a richer overall description of the leaf.…”

“…The number of their patterns are counted and the resultant histogram is taken as the descriptor of the entire leaf. Hierarchical string cuts (HSC) [9] characterizes a contour segment using the spatial distribution information of the contour points relative to its string that cuts the segment off the whole contour. This method is very fast and suitable for large shape database retrieval with a highly competitive accuracy.…”

“…It can be seen from Table 1 that all the nine competing methods consistently achieved significant improvements on their identification accuracies. The identification accuracies of HSC [9], MDM-CD-RM [16], MDM-ID-RA [16], IDSC [6], SPTC [6], BCF+SVM [50], BCF [50] To further observe the detailed behaviour of the proposed method in comparison with the benchmark methods in classifying soybean cultivars, we present their intermediate results of the dissimilarity measures in a visualization form. Fig.…”

Affine Invariant Point-Set Matching Using Convex Hull Bisection

“…The first dataset is MPEG-7 Core Experiment CE-Shape-1 (MPEG-7 CE-Shape-1) dataset [27] as shown in Figure 8. As Wang and Gao suggested this database was widely used to estimate the performance of shape recognition methods [28]. This dataset contains 70 different classes.…”