Adaptive quantization in multispectral image compression for equalizing visual error distribution

This article proposes an adaptive nonlinear quantization method for multispectral image compression. When linear scalar quantization is applied for multispectral image compression, extremely large error is perceived in low-luminance colors due to the nonlinear phenomenon of human vision. In the proposed method, quantization tables are switched pixel by pixel depending on the corresponding luminance. The switching rule is determined according to the relationship between the luminance and the error in the uniform color space. As a result, distribution of the error in the uniform color space can be equalized and the error in the low-luminance pixels is suppressed. Experimental results using a 16-band multispectral image of an oil painting shows the effectiveness of the proposed method.