8 Colour Quantization of Colour Construct Code in CIELAB Colour Space Using K-Means Clustering and Hungarian Assignment

摘要

Color Construct Code (CCC) has been introduced to store data with larger size by applying colour feature as compared to traditional 2D QR code. To decode CCC, colour quantization and colour correction process in pre-decoding process are essentially important in order to obtain data. Unfortunately, colour feature in CCC is always easily influenced by illumination. Meanwhile, colour space i.e. RGB and HSV is also device-dependent. Our work addresses these two issues and introduces a method to quantize and correct the 8 colour of CCC based on its CIELAB colour space. We combined K-mean clustering for the colour reduction and Hungarian assignment for the colour correction purpose. The results showed that our method does not only manage to correct the colour under varying illumination, but it is also independent to the device.