Block-Based Connected-Component Labeling Algorithm Using Binary Decision Trees

Abstract

In this paper, we propose a fast labeling algorithm based on block-based concepts. Because the number of memory access points directly affects the time consumption of the labeling algorithms, the aim of the proposed algorithm is to minimize neighborhood operations. Our algorithm utilizes a block-based view and correlates a raster scan to select the necessary pixels generated by a block-based scan mask. We analyze the advantages of a sequential raster scan for the block-based scan mask, and integrate the block-connected relationships using two different procedures with binary decision trees to reduce unnecessary memory access. This greatly simplifies the pixel locations of the block-based scan mask. Furthermore, our algorithm significantly reduces the number of leaf nodes and depth levels required in the binary decision tree. We analyze the labeling performance of the proposed algorithm alongside that of other labeling algorithms using high-resolution images and foreground images. The experimental results from synthetic and real image datasets demonstrate that the proposed algorithm is faster than other methods.

YACCLAB - Yet Another Connected Components Labeling Benchmark

“…In the following, we use acronyms to refer to the available algorithms: CT is the Contour Tracing approach by Fu Chang et al [21], CCIT is the algorithm by Wan-Yu Chang et al [9], DiStefano is the algorithm in [20], BBDT is the Block Based with Decision Trees algorithm by Grana et al [6], LSL STD is the Light Speed Labeling algorithm by Lacassagne et al [19], SAUF is the Scan Array Union Find algorithm by Wu et al [16], which is the algorithm currently included in OpenCV.…”

“…A set of synthetic random noise images has been used in many papers [4], [5], [6]. We have selected and included in YACCLAB the publicly available set of [6], because it is the only one already published and already used in other works [7], [8], [9]. This allows the researchers to easily compare their results with the literature.…”

“…The final decision tree is generated by another program, so a caveat to the reader is mandatory: the code is definitely ugly! Another variation of Block Based analysis was proposed in [9], which is reported to be faster than the previous one. We also include this algorithm thanks to the availability of the source code on the authors web pages, making the signature compliant with our standards.…”

Optimized Connected Components Labeling with Pixel Prediction

“…Another variation of Block Based analysis was proposed in [4], which is reported to be faster than the previous one. We also include this algorithm thanks to the availability of the source code on the authors web pages, making the signature compliant with our testing standard.…”

“…In the following, we use acronyms to refer to the available algorithms: CT is the Contour Tracing approach by Fu Chang et al [3], CCIT is the algorithm by Wan-Yu Chang et al [4], DiStefano is the algorithm in [5], BBDT is the Block Based with Decision Trees algorithm by Grana et al [7], LSL STD is the Light Speed Labeling algorithm by Lacassagne et al [12], SAUF is the Scan Array Union Find algorithm by Wu et al [16], which is the algorithm currently included in OpenCV, CTB is the Configuration-Transition-Based approach described in [10]. Our method is denoted as PRED.…”