A Systematic Analysis of Transform Coefficients and Block Decomposition for Texture Enhancement With Orthogonal Polynomials Model
提出一种三阶段纹理增强技术，通过分析正交多项式变换系数的方向特性，对图像块进行分解并分别调整亮度和锐化纹理，实验在标准图像集上验证效果优于现有方法。
Although numerous image enhancement techniques have been reported in the literature, enhancing a specific feature of an image is a challenging task and requires careful study. With that said, a novel threefold texture enhancement technique in the orthogonal polynomials transformed domain is proposed in this paper, by identifying the contribution of coefficients toward particular low-level characteristics. In the first fold, a systematic semantic analysis of orthogonal polynomials transformed coefficients is carried out in terms of the response of the basis operators and the directional properties of the coefficients are identified. A block decomposition scheme is then proposed with Meyer's model and blocks with different low-level characteristics are identified. In the second fold, an adaptive brightness adjustment scheme is proposed for nontextured blocks. In the case of texture blocks, besides brightness adjustments, a texture sharpening scheme based on the modification of the ac transform coefficients is proposed as the third fold. The proposed texture enhancement scheme is experimented on the standard benchmark University of Illinois at Urbana-Champaign images. The performance of the proposed texture enhancement scheme is measured with standard measures and compared with existing schemes. The results are encouraging.