图像处理知识点复习.doc

The ic Spectrum 在实际的图像处理应用中,最主要的图像来源于电磁辐射成像。电磁波谱频率从高到低(波长从短到长)分别为伽马射线、 X 射线、紫外线、可见光、红外线、微波、无线电波。可见光只是电磁波谱中一个很小的部分。不同的电磁辐射波有各自的成像方式,其应用领域也不尽相同。 Non-Electro-ic Imaging Modalities ( modality [m ?u?d? liti] 模式) 非电磁成像模式 Acoustic imaging [ ?? ku:stik i?m?d?i?] 声波成像 A Simple Image Formation Model (简单的)图像形成模型静止的、单色的、平面的图像可定义为一个二维函数),(yxf ,其中, ),(yx 是空间坐标,而在坐标),(yx 处的幅值 f 称为图像在该点处的强度或灰度。),(),(),(yxryxiyxf?),(yxf 可以看成由两个分量组成: (1) 照射分量),(yxi :所见场景的入射光量,大于零且为有限值, 由光源确定。(2) 反射分量),(yxr :场景中物体反射光量的能力,即反射率,在 0 (全吸收) 和 1 (全反射)之间, 由场景中物体的特性确定。 We denote images by two-dimensional functions of the form),(yxf .),(yxf is nonzero and finite. The function ),(yxf may be characterized by ponents: (1) ponent ),(yxi : The amount of source illumination incident on the scene being viewed. T he nature of),(yxi is determined by the illumination source. (2) ponent ),(yxr : the amount of illumination reflected by the objects in the scene. ),(yxr is determined by the characteristics of the imaged objects. reflectance is bounded by0 (absorption) and 1 (total reflectance) Image Sampling and Quantization 数字图像的质量在很大程度上取决于采样和量化中所用的采样数和灰度级。空间分辨率是图像中可辨别的最小细节,采样数是决定一幅图像空间分辨率的主要参数。灰度分辨率是指在灰度级别中可分辨的最小变化, 由量化的等级决定, 灰度级通常是 2 的整数次幂。 The output of most sensors isa continuous voltage waveform. To create a digital image, we need to convert the continuous sensed data into digital form. This involves two processes: sampling and quantization. The quality ofa digital image is determined toa large degree by the number of samples and discrete gray levels used in sampling and quantization. Spatial resolution is the smallest discernible [d ??s?:n?b?l] detail in an image. Gray-level resolution refers to the smallest discernible change in gray level. 保持图像的灰度级不变( 256 ), 逐渐降低图像的空间分辨率( 512 × 512 至 16× 16,从(a) 到(f) 采样数减少)。(注:为便于比较,所有的小图像都放大到了原图像的大小) 图(b) 与图(a) 基本相似;图(c) 中,帽檐处出现锯齿状,且头发有变粗的感觉;图(d) 中, 棋盘效应非常明显;图(e) 已几乎不能分辨出人脸;图(f) 单独观看完全不知其中为何物。总结:空间分辨率高的图像包含的像素多,图像清晰;空间分辨率低的图像,图像质量差,严重时产生频率混叠,画面出现棋盘效应( checker board effect )。对于同样的采