Novel methodology for construction and decoding of color data codes

Abstract

Colors can be employed as efficient sources of data conveyance as all their scalar descriptive attributes are transferred in the same spatial component. Although there is a great deal of ongoing research on each aspect of the process, palette construction (which establishes the basis of the whole practice) and decoding (which is the ultimate goal) issues deserve extra emphasis. The term palette herein states the combination of specific colors to be included in the symbology, disregarding the permutation order. That is, the number of colors to be used is the main determinant of the conveyance capacity, while the selection of colors strongly affects the decoding reliability. These two issues constitute the main dealings of this dissertation. In this context, 3D barcodes have been chosen as the exemplary application area, mainly because of their ubiquity and relatively easier implementation. The term 3D or 3-Dimensional describes herein a barcode utilizing more than two colors. As a machine-readable representation method of data, barcode technology provides one of the most recognized and cost-effective solutions to the quest for dispatching information with higher density and consistency. Hence, introducing colors to barcodes has been an attractive field of research for years, mainly with the motive to increase information density. A significant amount of work attempts to design a variety of 3D barcodes; nevertheless, only a few focalize on the issue of construction of the palette in particular. Even the existing ones appear either suboptimal or specific rather than generic. Although increasing the number of colors will directly leverage the capacity, in the presence of various disruptive effects it would further hamper distinguishability. In this respect, the palette structure and decoding approach should be in close accordance, considering the estimated characteristics of the anticipated distortion. This conception points out the main axes of this dissertation. In this direction, first, the peculiarities of a generic 3D-barcoding practice and the properties of field-specific interference were identified. In line with the derived operational barcode color acquisition model, the need for a plausible methodology for constructing robust color palettes was thus addressed. Along with, relevant techniques and metrics were devised to evaluate and compare pallets. Furthermore, a novel decoding schema has been developed as the consequential objective. It is acknowledged that selecting colors in the reference space as farthest apart as possible improves the expected performance of authentication. Therefore, the palette construction issue was worked out here as the inverse formulation of the well-known sphere-packing problem. The proposed methodology, as well as the comparison metrics (such as the accuracy-requirement cost, efficiency, cost-effectiveness, or palette quality), can be applied axiomatically to any number of colors. Also, the approach taken is independent of the coding system, operational conditions, or devices so that it can be generalized for the favored color space. Further, the 'iterative decoding with predictive convergence' (IDPC) method developed herein presents a novel concatenated scheme, which was not fully available in the congener methods used in 3D barcodes. As a derivative of the iterative decoding, the convergence step of the IDPC tries to reach the local error minima estimate by establishing a coaction between the inner and outer layers of the schema. To achieve optimal iteration implementation, the uncertainty measure presented here was utilized, which seems to be suitable to exploit in diverse congruent areas as well. Numerous palettes (of 3 ≤ N ≤ 100) were designed, analyzed, and compared with some of the known counterparts, along with the study. Results show that constructing more efficient palettes can be realized through the proposed methodology. The methods developed were further tested and evaluated through computer simulations and real-life experiments. Simulations were conducted mainly to examine the influence of the palette selection on color authentication, as well as the contribution of the proposed predictive convergence to the decoding performance. In this framework, a novel interleaving technique was also devised and applied. In addition to the simulations, comprehensive experiments were performed in different environments. The observed average performance of the suggested decoding method was over 99% in the tests conducted; besides it was confirmed that exercising the convergence provided the resolution of most (i.e. over 84%) of the errors remaining from the previous step. We can propound that, although the importance of color selection was emphasized in several previous studies, the need for a generic methodology proposal for robust palette construction, moreover, explicit metrics for evaluating and comparing palettes were not thoroughly afforded. This dissertation aims to contribute to the efforts to fulfill this highlighted need. Furthermore, the proposed decoding algorithm promises better performances through the revealed convergence implementation. It is also envisaged that the depicted concepts, methods, and findings would be instrumental as well for the other application areas of using colors for information representation and transfer, beyond 3D barcodes.