Abstract
The article deals with the modification of machine learning methods for embedding in an digital image compression framework. The modification leads to the need to change both the structure of machine learning methods and the algorithms for tuning the methods. When tuning machine learning methods, we use a quality score that is relevant to the problem of digital image compression. We develop an algorithm for tuning binary classifiers, and then we generalize the algorithm to the situation of tuning arbitrary machine learning algorithms when compressing digital images. We describe the data structures, numeric arrays, and calculation formulas used in the tuning algorithm of the modified machine learning method. We embed the modified machine learning based method into a digital image compression framework. We perform computational experiments to study the effectiveness of modified machine learning methods for natural image compression. Computational experiments confirm the high efficiency of modified machine learning methods for digital image compression.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have