压缩映射

ABSTRACT Fractal image coding, which is also called attractor image coding, is a emergent method of image compression during the last decad...

分形图像编码方法是近十年来诞生并发展起来的一种新型图像压缩方法，它将图像编码为一组收缩映射，由这组收缩映射的不动点近似待编码对象。

Based on the summarizing of the exiting algorithms of the texture mapping and antialiasing, this paper goes deeply into the basic theory and processing of texture mapping. The texture compress algorithm which can satisfy the need of the memory and time is given for the stochastic sampling in the inverse texture mapping; the geometry transformation in the texture mapping are discussed and the concepts of the unitary parametrization and the patch parametrization is proposed, some kinds of the unitary parametrization methods of parametric surfaces and the geometry transformation in the patch parametrization are also presented; in order to solve the aliasing in the parametrization of intermediate surface-sphere in two-part texture mapping, an area-preserving transformation is presented also; to improve the quality of the graph the aliasing in the texture mapping is analyzed from the aspect of signal process, and the modified summed-area tables algorithms is proposed; the distortion of the scanline conversion algorithms under perspective transformation is analyzed and the proper interpolation algorithm is presented so the aliasing problem of the polyhedral object after the unitary parametrization is efficiently solved.

本文在总结现有的纹理映射和反走样算法的基础上，深入研究了纹理映射的基本原理与过程，根据反向纹理映射对纹理随机采样的特殊要求，提出了一种可满足存储空间和寻址时间的纹理压缩存储算法；通过对纹理映射中几何变换理论的进一步深入探讨，提出了整体参数化和面片参数化的概念，并给出了参数化曲面的各种整体参数化方法和面片参数化的几何变换形式；针对现有的两步映射算法所存在的中间曲面—球面的参数化问题，提出一种等积映射方法，较好地实现了由非参数化面片拼接而成的多面体表面的纹理映射；从信号采样的角度深入分析了纹理映射中走样产生的原因，提出了改进的区域求和表算法，使图形生成的效果有所提高；还分析了一般的扫描线算法在透视变换中所存在的问题，提出了一种正确的插值算法，从而解决了整体参数化后多面体表面纹理映射的几何变形问题。

Based on contraction mapping theorem of analysis mathematics, the non-steady flow mathematical model of mine ventilation network was analyzed.

利用分析数学的压缩映射原理，对矿井通风网络非稳定流动数学模型进行了分析，得出了数学模型数值解收敛的条件，确定了以惯性系数排序选择最小生成树的回路选择方案，为矿井通风网络非稳定流动数学模型数值解的收敛性提供了理论依据。

Topics covered include: nonlinear ordinary differential equations; planar autonomous systems; fundamental theory: Picard iteration, contraction mapping theorem, and Bellman-Gronwall lemma; stability of equilibria by Lyapunov's first and second methods; feedback linearization; and application to nonlinear circuits and control systems.

课程的主题包括：非线性常微分方程组；二维自治系统；与课程相关的基本理论，如比卡迭代法，压缩映射定理，Bellman-Gronwall引理，平衡点稳定性分析的李雅普诺夫第一种方法和第二种方法，反馈的线性化；以及非线性电路和控制系统的实际应用。

The second section: Under the conditions of nonlinear boundary controbility, we consider the initial boundary value problem of Camassa—Holm equations with dissipative. By using the contractive mapping fixed point theorem and a priori estimates, the existence of global smooth solution, global attractor in H〓, time periodic solution or almost-periodic solution and the global exponential stability are proved.

第二部分：在非线性控制边界条件之下，对于带耗散项的Camassa—Holm方程的初边值问题，用压缩映射不动点原理及先验估计方法，证明了整体光滑解的存在性、整体解的指数稳定性、H〓空间中整体吸引子的存在性以及时间周期解和殆时间周期解的存在性。

This paper uses contraction mapping approach in order to give sufficient condition for solvability about this kind of matrix equation in given conditions.

文章利用压缩映射思想给出了一类矩阵方程在给定条件下有解的充分条件。

Firstly,the wavelet-based contourlet coding using an SPIHT-like algorithm is utilized on the plain image,and as a result,the compressed datum stream could be achieved which is then mapped to a 3D bit matrix. Afterward,a chaotic sequence is generated by Lorenz chaotic map,and it is preprocessed to a bit sequence which is used to permute and substitute the elements of the 3D bit matrix achieved above. In the end,the processed 3D matrix is mapped back to datum stream,and the encrypted compressed image could be achieved if decoding and inverse transform are performed on the datum stream.

首先采用基于小波的Contourlet变换的类等级树集合分割编码算法对明文图像进行压缩，得到压缩数据流，然后将压缩数据流映射为一个三维位矩阵；利用Lorenz混沌映射产生混沌序列，并对其进行预处理得到比特值序列，根据比特值序列对上述三维位矩阵进行置乱和替代操作；将置乱和替代后的位矩阵重新映射为数据流，并对其进行解码和反变换操作，得到加密后的压缩图像。

As the inversion and compression attack includes the question relating to solving thepreimage from image of function, a subdivision algorithm for solving preimage of function withbounded derivative based on Lagrange mean value theorem is proposed firstly.

由于逆推压缩攻击方法涉及混沌映射的求逆，针对复杂混沌映射求逆困难的问题，运用拉格朗日中值定理，我们提出了求解导函数有界的函数原象的分割算法，给出了使分割算法计算复杂性上界达到最小的充分条件。

Some sufficient conditions for existence and uniqueness of periodic solution to Duffing equation are obtained using trigonometrical series and contraction mapping.

利用三角级数理论及压缩映射原理，研究时滞Duffing型微分方程，给出此类方程存在唯一周期解的几个充分条件。

According to contraction theory, this algorithm is based on gradient descent method, which adopts linear average as performance evaluation of value function. So the iterative process of value function becomes a convergent process to a fixed value.

该算法基于梯度下降法，根据压缩映射原理，通过采用线性平均法作为值函数估计的性能衡量标准，把值函数估计的迭代过程转化为一个收敛于不动点的过程。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。