Rapid autofocusing via 1 bit transform for real-time system to detect mycobacterium tuberculosis

Autofocusing is a key building block in the modern automated high-throughput systems for a quantitative analysis of microscopy images. For that reason, it has received great attention, and diverse autofocus algorithms have been proposed in the literature. A fast autofocus algorithm is presented in order to build a real-time system for diagnosing the sputum smear samples. The method involves transforming the sputum smear fluorescence images from a multibit to a 1 bit/pixel representation. This results in a substantial reduction in the computational cost, while maintaining a good focusing performance. Experimental results are presented to validate the feasibility and the effectiveness of the proposed method.