Full search content independent block matching based on the fast fourier transform

FULL SEARCH CONTENT INDEPENDENT BLOCK MATCHING BASED ON THE

FAST FOURIER TRANSFORM

Steven L. Kilthau, Mark S. Drew, and Torsten M?ller School of Computing Science, Simon Fraser University,

Vancouver, B.C., Canada V5A 1S6

{kilthau,mark,torsten}@cs.sfu.ca

ABSTRACT

In this paper, we present a new algorithm for solving the block matching problem which is independent of image content and is faster than other full-search methods. The method employs a novel data structure called the Windowed-Sum-Squared-Table , and uses the fast Fourier transform (FFT) in its computation of the sum squared difference (SSD) metric. Use of the SSD metric allows for higher peak signal to noise ratios than other fast block matching algorithms which require the sum of absolute difference (SAD) metric. However, because of the complex floating point and integer math used in our computation of the SSD metric, our method is aimed at software implementations only. Test results show that our method has a running time 13%-29% of that for the exhaustive search, depending on the size of the search range.

1. INTRODUCTION

The block matching problem is one that occurs in many areas of the image processing, multimedia, and vision fields. In this paper, our focus is on the application of block matching to the computation of motion vectors for video compression. Because of their widespread use, many motion vector compensation algorithms have been developed and are in use today.

Block matching algorithms all attempt to minimize some measure of similarity between a template block of pixels in the current image to all candidate blocks in the reference image within a given search range. The two most popular similarity measures used are the sum of absolute difference (SAD) and the sum of squared difference (SSD). Given a block size B and a displacement vector ()v u , for a candidate block relative to the template block, the metrics are defined as:

()()

∑∑?=?=?++?=

101

1

),(,,B j B i t t

v u v j u i f

j i f SAD (1)

()()()∑∑?=?=?++?=

101

21

),(,,B j B i t t

v u v j u i f

j i f SSD (2)

Because of its lack of multiplications the SAD metric is far more convenient for use in hardware designs, and is therefore used almost exclusively. However, minimizing the SSD metric corresponds to maximizing the peak signal to noise ratio (PSNR), whereas minimizing the SAD metric does not. Therefore, if a maximum PSNR is desired, SSD should be the metric of choice.

All existing block matching algorithms can be roughly grouped into two categories. The first category consists of those algorithms that are not guaranteed to find the best matching block within a given search range, but instead use a heuristic approach to guide the search. These methods examine only a subset of the possible locations within the search range, and hence can be computed very efficiently. Some of the most popular methods are the three step search [8], the two dimensional logarithmic search [6], and their many successful variants such as the one found in [3]. Because of their speed, these suboptimal methods are of great interest. However, they are prone to getting trapped in local minima and thus are not appropriate for applications which require a maximum PSNR.

The second category, and the subject of our focus, consists of those algorithms which are guaranteed to find the optimal matching block within a given search range. In recent years, many algorithms have been developed for this type of search [2, 5, 9, 10, 11]. All of the algorithms in this category achieve their speedup through early elimination of candidate search positions; however they suffer from the fact that their performance depends largely on the content of the image sequence being encoded. Much of the recent research [2, 9, 11] eliminates search positions through application of the Minkowski Inequality:

p

i

p

i

p

i p

i

p

p

i

i

i b

a

b a ∑∑∑

+

≤

+ (3)

For the case 1=p , substitutions i i i y x a ?=, and i i y b = yield the following common form:

∑∑∑?≥?i

i

i

i

i

i

i

y

x y x . (4)

Many algorithms make use of a pyramid version of Eq.4 for early elimination of candidates. Notice however that for 2=p , the Minkowski Inequality requires the computation of the square root. Since the square root operation is extremely expensive, techniques that rely on the Minkowski Inequality to eliminate search positions can not efficiently use the SSD metric, and are instead only able to use the SAD metric. Hence, all methods that require the Minkowski Inequality cannot guarantee a maximum PSNR value, while still maintaining computational efficiency.

One technique that doesn’t fit into the above categorization is the phase-correlation method [1]. This technique works by computing the cross-correlation of the template block with the corresponding search range and identifying a set of candidate correlation peaks. The algorithm then evaluates a difference measure at those points and chooses the minimum as the solution to the block matching problem. Although this algorithm cleverly reduces the computational complexity it has been shown to

identify spurious solutions, and as such is not guaranteed to maximize PSNR.

Since our method uses the SSD metric to find the minimum matching block within the given search range, we are guaranteed to find the blocks that maximize the PSNR of the predicted image. Furthermore, our method achieves its speedup regardless of the content of the image sequence being encoded. The details of our algorithm are given in §2, experimental results are given in §3, and we discuss conclusions in §4.

2. THE FFT BLOCK MATCHING ALGORITHM

In order to maximize the PSNR, our algorithm minimizes the SSD metric given in Eq.2. Following a trivial expansion, the mathematical definition of our per-block computation is given by:

[()()()()]∑∑?=?=???++?+++1

01

01

2

12,,,2,,min B j B i t t t t v u v j u i f j i f v j u i f j i f (5)

Since the term ()2,j i f t appears across the entire minimum, it

can be removed from the sum without affecting the resulting solution. Removing this term and separating the sum leaves us with the following equation:

()()()

∑∑

∑∑?=?=??=?=??++?++10

1

1101

02

1

,,,2,min B j B i t t B j B i t v

u v j u i f j i f v j u i f (6)

The FFT Block Matching Algorithm (FFTBMA) that we propose

computes Eq.6 using three basic steps:

1. Resize input image to include a zero pad

2. Compute the windowed sum squared table

3. Compute a per-block convolution sum

Step 1 is simply to allow convenient calculation of the SSD metric without using conditionals for those search locations that lie outside of the dimensions of the original image. Given a search range of P ± we apply a zero pad of P pixels around the entire image. This simple preprocess eliminates the need for conditionals within the innermost loops of our algorithm and greatly increases its speed. Similarly, this also improves the performance of the exhaustive search, and as such is used in our implementation of that algorithm as well. For convenience, we assume here that the original dimensions of the image are a multiple of the block size, B. If this is not initially true, the dimensions of the image are increased to compensate for this prior to the application of the zero pad.

Steps 2 and 3 are discussed in §2.1 and §2.2 respectively.

2.1 Windowed Sum Squared Table (WSST)

To compute the first term of Eq.6, we use a variant of the well known summed area table (SAT), introduced in [4]. Given an input image f , a summed area table is a new image SAT f such that

()()∑∑

≤≤=

i

k j

l SAT l k f j i f ,, (7)

Summed area tables can be very easily computed by applying the following recurrence, being careful to set ()j i f , to zero when either of the indices is negative:

()()()

[()()]

1,11,,1,,????+?+=j i f j i f j i f j i f j i f SAT SAT SAT SAT (8)

The WSST differs from the SAT in that each pixel needs to represent a sum of squares , where the sum extends only over the last B B × sub-image (window), with B the block size.

Our approach to creating the windowed sum squared table consists of two steps. In the first step we compute a sum squared table (SST), and in the second step we confine the sum to the last B B × sub-image. Using a variant of Eq.8 would imply that for an 8 bit image of size H W × we may need to store values as large as WH 2255. For large video streams such as those used in HDTV, this can easily exceed the maximum value representable by a 32 bit integer. So although this algorithm seems to follow directly from the recurrence of Eq.8, care is needed to prevent overflow of intermediate calculations.

To solve the overflow problem, we initially divide the image into blocks of size B B × and compute an SST over each block. By assumption, the image size is divisible by B , and we assume that 22255B can be represented with a 32 bit integer. The result of this computation is a new image where each block constitutes a sum squared table defined by the following recurrence:

()[()()

()()]

1,11,,1,,2????+?+=j i f j i f j i f j i f j i f SST SST SST SST (9)

We now need to combine the individual SST’s to create the final Windowed SST. For this we note that the sum of squares over any rectangle confined to a single block, with lower left corner ()j i , and upper right corner ()l k , is given by:

()()()()j i f j k f l i f l k f

t s f SST SST SST SST

j l s i

k

t ,,,,),(2

+??=∑∑== (10)

Using Eq.10 we can easily derive the sum of squares (SS) for an arbitrary B B × region as an SS over rectangles in 4 neighboring blocks. By using coordinates that are local to the block containing the upper right corner we arrive at the following equation:

()()()[]

()()[]()()[]()()[()()]

B j f B i f B j B i f f B j i f i f j B i f j f j i f t s f j i f SST SST SST SST SST SST SST SST SST j B j s i

B i t WSST ????????+??+???+???+==

∑

∑

+?=+?=,11,,1,1,1,,,1,,,1

1

2

(11) Fig.1 depicts Eq.11 visually. By applying this equation for every pixel we complete the computation of the windowed sum squared table for the entire frame.

2.2 Per-Block Convolution Sum

In this section we show that computation of the second term in Eq.6 amounts to the evaluation of a correlation sum for each template block, which we evaluate as a convolution sum. For a single candidate block, the second term of Eq.6 is just a dot product with the template block. However, computing this dot product for each of the ()212+P candidate blocks in the search range amounts to a correlation of the template block with the ()()B P B P +×+22 region corresponding to the square containing all pixels of all blocks in the search range. In order to efficiently compute the correlation, we will first convert it to a

B

)0,0(SST f

),(j i f SST

B

Figure 1. Computation of WSST relative to multiple SST’s

convolution and then use the fast Fourier transform (FFT).

For each template block, we create two images of size ()()B P B P +×+22. The first image, template , corresponds to the template block and is computed by simply multiplying the block by 2, reversing the pixels, and zero padding to the correct size. The pixel reversal effectively changes the correlation sum into an equivalent convolution sum. The second image, candidates , corresponds to the square containing all pixels of all candidate blocks in the search range. This square can be copied directly from the reference image. Given these two images, we can compute a new image, result , according to the following formula:

()()()candidates FFT template FFT FFT result ?=?1 (12) Since the convolution that we have performed is cyclic, the first 1?B rows and columns of result will contain wrap-around data that should be discarded. This leaves us with a usable portion of the image of size ()()1212+×+P P . Notice that this corresponds

to one solution for each of our ()212+P search locations, and is exactly what we desire.

Given that we have previously computed the windowed sum squared table, we can now easily find the minimum. Assuming that our template block is at offset ()y x , we simply perform a linear pass over the result image, evaluating the minimum matching block according to the following formula:

()()),(,min ,j i result P j y P i x f WSST j i ??+?+ (13) where []12,1,?+?∈B P B j i . It is important to note that ),(j i result must be rounded to the nearest integer before it can be combined with a value from the windowed sum squared table. As a last step, we simply need to correct the offset to account for the cyclic nature of the convolution. Given that Eq.13 identifies the pair ()j i , as the location of the minimum match, the resulting motion vector corresponding to the minimum matching block is then given by:

()()1,1,+??+??=B P j B P i mvj mvi (14)

2.3 Running Time Analysis

In our discussion of running time, we will assume without loss of generality that the dimension of the original image is N N ×, where N is divisible by B .

We will first develop an expression for the running time of the exhaustive search. Since full search algorithms are generally content dependent, there are certain cases where they will all

exhibit the same worst case running time. For each block, the exhaustive search requires the computation of the SAD or SSD metric at ()212+P locations. This results in an ()()

2212+ΟP B algorithm for each block. There are a total of 22B N blocks, so the exhaustive search has a total per-frame running time of: ()()

2212+ΟP N (15)

The FFT block matching algorithm that we have presented consists of two steps. In the first step we construct the windowed sum squared table. Since this amounts to only two linear passes

over the entire image, this step takes only ()

22N Ο per frame. In the second step we perform a per-block convolution. Each convolution consists of three applications of a

()()B P B P +×+22 real FFT, as well as ()222

B P + complex

multiplications. Therefore this step has a running time of ()()()

B P B P ++Ο2log 222 per block. A precise statement of the

total running time of the FFT block matching algorithm is then:

()

()()()()()

2232log 22222

2

+++++ΟB B P B P B P N (16)

For practical scenarios, asymptotic analysis is of no interest. To determine the cases for which our algorithm outperforms the exhaustive search, we analyze the following equation:

()()()()()()2232log 21222222+++++≥+B B P B P B P C P (17) We can safely assume that []4,12∈=B P R . Substituting R

into Eq.17 and simplifying results in the following expressing:

()()[]()31log 1122222++++≥B R R R C

B (18)

In §3 we show that the FFT block matching algorithm outperforms the exhaustive search for all relevant block sizes and a range of C , including values that characterize inefficient FFT implementations.

3. EXPERIMENTAL RESULTS

Our experiments test the performance of the FFT block matching algorithm against that of the exhaustive search. Both algorithms have been implemented using C/C++. For the FFT computation we use a publicly available package called the Fastest Fourier Transform in the West (FFTW) [7]. None of the code or libraries that we have used contain any machine specific instructions or assembler routines that would give either algorithm an unfair advantage. Our test platform is an AMD Athlon 900 MHz with 512 MB of RAM. All timings include reading the images, constructing the necessary data structures, and computing all motion vectors for each frame of the input image sequence.

Data Set Frames Width Height Motion Mother 199 176 144 slow Carphone 74 176 144 medium Football 160 360 240 fast

Table 1. Data sets used in the experiments.

Both algorithms have been tested on three data sets. The dimensions, number of frames, and type of motion classification of each are given in Table 1. The motion classification is shown as slow, medium, or fast, and indicates the degree of motion contained within the image sequence. For all tests, a block size of 16 is used, and only the luminance channel is considered.

The performance of each algorithm is measured using

Individual sum-squared-

execution time, PSNR, and the number of errors, E. The number of errors needs to be considered because although the FFT block matching algorithm is mathematically exact, the round-off error produced by the many computations used in the convolution sum will infrequently lead to a non-optimal match. We have observed that this occurs less than 0.5% of the time in practice, and when the result is non-optimal, the computed motion vector is usually matched to the next most optimal block. All data sets were tested using multiple values of the search range, P .

Exhaustive Search FFT BM Algorithm

P Time PSNR E Time PSNR E +/-8 14.06 38.20 0 4.01 38.20 74 +/-16 52.13 38.31 0 10.89 38.31 86 +/-24 114.51 38.34 0 15.19 38.34 52 +/-32 201.19 38.36 0 33.03 38.36 74

Table 2. Test results for mother image sequence.

Exhaustive Search FFT BM Algorithm

P Time PSNR E Time PSNR E +/-8 5.19 32.00 0 1.49 32.00 27 +/-16 19.21 32.02 0 4.01 32.02 36 +/-24 42.22 32.02 0 5.59 32.02 31 +/-32 74.18 32.03 0 12.36 32.03 25

Table 3. Test results for carphone image sequence.

Exhaustive Search FFT BM Algorithm

P Time PSNR E Time PSNR E +/-8 36.91 23.13 0 10.52 23.13 3 +/-16 136.65 23.46 0 28.38 23.46 2 +/-24 300.14 23.55 0 40.14 23.55 0 +/-32 527.42 23.60 0 87.10

23.60 2

Table 4. Test results for football image sequence.

As Tables 2, 3, and 4 show, the FFT block matching algorithm greatly outperforms the exhaustive search while still obtaining the same PSNR value, with running times of 13%-29% of that of the exhaustive search, dependent on P . Fig.2a shows a plot of the timings for the football image sequence, as well as the analytic curves in the inequality of Eq.17 (here we used 75.2=C ). In Fig.2b, we have solved Eq.18 for B , varying C and R . Since we have measured C to be between 2.2 and 3.0 in our implementation, Fig.2c shows a slice through the plot in Fig.2b for 3=C . With such a value of C , it is clear that the FFT block matching algorithm convincingly outperforms the exhaustive search for our previously defined range of []4,1∈R . Even for much less efficient implementations of the FFT, it is clear that our algorithm is faster.

In our tests, using 24=P consistently gives the best running times. This occurs because P is relatively large and the dimension of the FFT is then 6464×, which runs extremely fast because it is a power of 2. As we have mentioned, the FFT block matching algorithm occasionally fails to identify the best matching block due to round-off error, but as the PSNR indicates, the effect of this error is negligible.

4. CONCLUSIONS

Our new FFT-based block matching algorithm employs a novel data structure, the Windowed-Sum-Squared-Table, and exploits the FFT in its computation of the SSD metric. Because it is independent of image content, our algorithm runs faster than

Figure 2. (a) Timings for football image sequence.

(b) Solutions of Eq.18. (c) Slice through Fig.2b at 3=C .

existing full search algorithms, with speeds of 13%-29% of the

exhaustive search in practice. The FFT block matching algorithm is not heuristic-based and thus can consistently identify the best matching blocks, maximizing PSNR. The algorithm is well suited for software implementations requiring very low bit rates.

5. REFERENCES

[1] C. D. Kuglin and D. C. Hines, “The phase correlation image alignment method,” in Proceedings of the 1975 IEEE International Conference on Systems, Man and Cybernetics , pp 163-165, 1975.

[2] C.-H. Lee and L.-H. Chen. “A fast motion estimation algorithm based on the block sum pyramid,” IEEE Transactions on Image Processing , 6(11):1587-1591, 1997.

[3] D. W. Zhang, I. Ahmad, M. Liou. “Adaptive motion search with elastic diamond for MPEG-4 video encoding,” Proceedings of International Conference on Image Processing , 2001.

[4] F. C. Crow, “Summed-area tables for texture mapping,” Computer Graphics (Proc. of Siggraph), 18(3):207-212, 1984.

[5] H.-C. Huang and Y.-P. Hung, “Adaptive early jump-out technique for fast motion estimation in video coding,” Graphical Models and Image Processing , 59(6):388-394, 1997.

[6] J. R. Jain and A. K. Jain. “Displacement measurement and its application in interframe image coding,” IEEE Transactions on Communications , COM-29(12):1799-1808, 1981.

[7] M. Frigo and S. Johnson, “FFTW: An adaptive software architecture for the FFT,” in Proceedings of the International Conference on Acoustics, Speech, and Signal Processing , vol. 3, pp 1381-1384, 1998. [8] T. Koga, K. Iinuma, A. Hirano, Y. Iijima, and T. Ishiguro. “Motion-compensated interframe coding for video conferencing,” Proceedings of National Telecommunications Conference , vol. 4, pp G5.3.1-G5.3.5, 1981.

[9] W. Li and E. Salari. “Successive elimination algorithm for motion estimation,” IEEE Transactions on Image Processing , 4(1):105-107, 1995.

[10] Y.-C. Lin and S.-C. Tai, “Fast full-search block-matching algorithm for motion-compensated video compression,” IEEE Transactions on Communications , 45(5):527-531, 1997.

[11] Y.-S. Chen, Y.-P. Hung, and C.-S. Fuh. “A fast block matching algorithm based on the winner-update strategy,” in Proceedings of the Fourth Asian Conference on Computer Vision , vol. 2, pp 977-982, 2000.

(b) (c)

(a)

ContentProvider和Uri详解222

https://www.wendangku.net/doc/4e14200293.html,/s/blog_49f62c350101hhhl.html Android四大组件是Activity, Service, Content Provider, Broadcast Receiver。 Activity作为程序界面，直接与用户交互 Service运行在后台，没有界面，完成特定的功能 ContentProvider维护应用数据，方便应用本身或其它应用访问 Broadcast Receiver提供异步广播消息接收机制，便于各应用/组件进行交互 通过AndroidManifest.xml, 可以看到一个应用使用了哪些组件： attribute的定义可以参考 https://www.wendangku.net/doc/4e14200293.html,/guide/topics/manifest/manifest-intro.html 下面重点探讨Content Provider的实现和使用。 二．什么是ContentProvider

Content Provider维护特定的应用数据，并可以让其它应用轻松访问该数据。对数据使 用者来说它是数据提供者。它提供统一的接口对数据进行操作，使用者不用关心数据到底是如何存储的以及数据类型到底是什么。也就是说，Content Provider作为数据提供者，提供了对外共享本地数据一种机制，使Android应用能方便地基于该机制进行数据访问。 为了便于管理和访问，每个Content Provider必须有唯一标示，用Uri表示。Uri 类似http url, 构成如下： content://authority/path 所有Content Provider的Uri必须以content://开头，这是Android规定的。 authority是个字符串，它由开发者自己定义，用于来唯一标示一个ContentProvider。系统会根据这个标示查找ContentProvider。 path也是字符串，表示要操作的数据。可根据自己的实现逻辑来指定：content://contacts/people表示要操作ContentProvider为contacts下的people表content://com.android.contacts/people/#表示要操作表people中特定id的行(记录)。content://downloads/download/10/name表示要操作id为10的行的name字段。content://downloads/download/*表示操作download表中的所有字段。 总之，#匹配一个数字字符串，*匹配一个文本字符串。 可以看出, 实现一个自定义的Content Provider，要基于系统提供的基类ContentProvider，需要实现6个接口。大部分接口就是类似数据库的数据操作接口，实际上Content Provider是需要创建数据库并对数据库进行操作的。完成实现之后，在

content

https://www.wendangku.net/doc/4e14200293.html,模板页的使用 1，创建模板页 1）首先我们先建立一个网站（图一）， 。 2）在此项目上添加一个模板页，在网站上“右击鼠标”，选择“添加新相”，找到“模板页”（图二），名称自己定，语言当然就不用说了，然后点击添加就可以了！

3）现在我们再仔细的看一下这个模板页到底有什么功能，到底有什么神秘之处！我们双击如图所示的MasterPage.master文件，查看其原文件，我们首先看到第一行就是“<%@ Master Language="C#" AutoEventWireup="true" CodeFile="MasterPage.master.cs" Inherits="MasterPage" %>”，这句话是模板页的声明，如果我们仔细一点的话就会发现，它的声明和普通页面的声明很相似，主要有一下几点：1、模板页是以master 开始，后缀是“.master”，而普通的页面是以page开始的，后缀是“.aspx”，再一个很大的区别就是：在模板页中存在一个“ContentPlaceHolder”控件，这个控件就是用来填充内容页的地方，而普通的页面是没有的！（图四）我们可以在模板页的地方添加各种固定的信息如页面的logo或者是声明等！现在我们为这个模板页添加一个内容页，来看一下效果。

具体操作如下：我们在内容页的位置右击鼠标，选择“添加内容页”（图六）就可以。此时，我们会看到内容页的源码区，及其简单，就三句话。

其中“MasterPageFile="~/MasterPage.master"”这句话是定义是模板页， 再看下一行“ ”， “ContentPlaceHolderID="ContentPlaceHolder1"”为我们指定了模板页中的可以添加内容页的ContentPlaceHolder，有时候我们的模板页会不只有一个ContentPlaceHolder，我们要根据自己的需要来选择！我们选择“设计”模式，来看一下效果，只有内容页的地方允许我们编辑，这就实现了模板页的作用！假如我们的内容页也想利用模板页的内容，我们可以点击ContentPlaceHolder右上角的那个小三角，然后选择“默认为模板页内容就可以了”（图八）！ 在母版页中，可以添加多个ContentPlaceHolder控件，通过在工具箱的标准栏中选择ContentPlaceHolder，拖动到母版页中。在ContentPlaceHolder区域中，也可以直接输入内容或添加https://www.wendangku.net/doc/4e14200293.html, 控件，ContentPlaceHolder区域中的内容将作为内容页的默认内容呈现。 2,使用模板页

药品基础知识大全

药品基础知识大全 药品 药品是指用于预防、治疗、诊断人的疾病，有目的地调节人的生理功能并规定有适应证、主治、用法、用量的物质。 中药饮片 是指在中医药理论的指导下,可直接用于调配或制剂的中药材及中药材的加工炮制品。 毒性药品 是指毒性剧烈，治疗量与中毒剂相近，使用不当会致人中毒或死亡的药品。 毒性中药管理的品种有27种按卫生部规定，它们是：砒石（红砒、白砒）、砒霜、水银、生马钱子、生川乌、生草乌、生白附子、生附子、生半夏、生南星、生巴豆、斑蝥、青娘子、红娘子、生甘遂、生狼毒、藤黄、生千金子、生天仙子、闹羊花、雪上一枝蒿、红粉、白降丹、蟾酥、轻粉、雄黄、洋金花。 医疗器械 单独或者组合使用于人体的仪器、设备、器具、材料或者其他物品，包括所需要的软件；其使用旨在达到下列预期目的：国家对医疗器械实行产品生产注册制度。有效期是4年。 消毒产品

消毒产品包括消毒剂、消毒器械、卫生用品和一次性使用医疗用品。消毒产品不是药品，其外包装、说明书、标签上不应出现或暗示对疾病有治疗效果。 保健食品 具有特定保健功能或者以补充维生素、矿物质为目的的食品。即适宜于特定人群食用，具有调节机体功能，不以治疗疾病为目的，并且对人体不产生任何急性、亚急性或者慢性危害的食品。 指以涂擦、喷洒或者其他类似的方法，散布于人体表面任何部位（皮肤、毛发、指甲、口唇等），以达到清洁、消除不良气味、护肤、美容和修饰目的的日用化学工业产品。 特殊用途化妆品 是指用于育发、染发、烫发、脱毛、美乳、健美、除臭、祛斑、防晒的化妆品。化妆品标签、小包装或者说明书上不得注有适应症，不得宣传疗效，不得使用医疗术语。 药品和保健品的区别 保健食品与药品最根本的区别就在于保健食品没有确切的治疗作用，不能用作与治疗疾病，只是具有保健功能，既不可宣传治疗功效。对某些保健食品利用非法广告进行夸大宣传，号称“包治百病”，我们一定要有清醒的认识，以免受到广告的欺骗耽误正常的治疗、加重病情。 药品本身的特殊性

pull短语及详细用法教学内容

pull away [ phrasal verb ] : to begin to move farther ahead in a race, competition, contest, etc. They pulled away in the second half and won the game easily. — often + from In the final lap, he pulled away from the other cars and won. pull back [ phrasal verb ] 1 : to decide not to do something that you had intended to do or started to do The buyers of the house pulled back [=pulled out] at the last minute. 2 pull back or pull back (someone or something)or pull (someone or something) back : to move back from a place or position or to cause (someone or something) to move back from a place or position The soldiers were outnumbered and were forced to pull back. [=retreat, withdraw] The general pulled his army back. 3 pull (something) back or pull back (something)Brit, sports : to score (a goal, point, etc.) so that you are not as far behind in a game as you were before They were behind 2?0 but they pulled back a goal [=they scored a goal to make the score 2?1] early in the second half. pull down [ phrasal verb ] 1 pull down (something) or pull (something) down 1 a : to move (something) down I pulled down the shade. He always wears his baseball cap pulled downover his eyes. 1 b : to destroy (a building) completely The wreckers pulled down [=demolished] the building. 1 c : to make (something) smaller in amount or number : to reduce or lower (something) The rumors that the company was filing for bankruptcy pulled stock prices down. 2 pull down (someone) or pull (someone) downUS, informal : to cause (someone) to become sad or depressed The loss really pulled the team down. [=brought the team down] 3 pull down (something) informal 3 a : to earn (a particular and usually large amount of money) He pulls down [=makes, pulls in] more than a million dollars a year. 3 b : to get (something) The show has pulled down high ratings. pull in [ phrasal verb ] 1 : to arrive at a place and come to a stop “When are our guests coming?” “I think they just pulled in.” The train pulled in on time.

临床常用药物大全

第一章.消化系统疾病常用药物 第一节抗酸药：是一类弱碱性物质，口服能中和胃酸 碳酸氢钠：0.5g/片 【适应病症】：治疗胃酸过多，代谢性酸中毒及高钾血症。 【用法】：餐前服用。 【注意事项】：静脉滴注时应防止渗漏，应注意给药速度，5％碳酸氢钠为高张性溶液，滴注过快会抑制心脏，使血压骤降，不利于心脏复苏；对血钾过低者不宜立即应用，忌与酸性药物配伍，除普鲁卡因胺外，不宜与其他常用的心肺复苏药物合用；口服易产生CO2，将要穿孔的溃疡患者忌用。 【不良反应】：引起继发性胃酸分泌过多。用量过大可致碱中毒。 铝碳酸镁(胃达喜)：500mg/片 【适应病症】：治疗消化性溃疡，胃炎。 【用法】：每次1～2片，3 次/日，餐后1～2小时或睡前咀嚼服用。 【注意事项】：大剂量服用可能有胃肠道不适，如消化不良和软糊状便，肾功能不全者长期服用应定期监测血中的铝含量，可影响四环素、环丙沙星、氧氟沙星的吸收。 【不良反应】：本药不良反应少而轻微，仅少数患者有胃肠道不适、消化不良、呕吐、大便次数增多或糊状大便，个别有腹泻。 磷酸铝（洁维乐凝胶）：20g/包 【适应病症】：治疗消化性溃疡，胃炎，食管炎，胃酸过多等 【用法】：用前先摇匀，挤出凝胶直接服用，也可就水服用，成人每天2次，每次1包【注意事项】：每袋磷酸铝凝胶含蔗糖2.7g，糖尿病患者使用本品时，不超过1袋。慢性肾功能衰竭患者禁用，高磷血症禁用。对卧床不起或老年患者，有时会有便秘现象，此时可采用灌肠法。 【不良反应】：可见恶心、呕吐、便秘、大剂量可致肠梗阻；长期服用可致骨软化、脑病、痴呆及小红细胞性贫血，本品可影响某些药物的吸收。 大黄苏打：0.3g/片 【适应病症】：有健胃、制酸作用，用于食欲不振、消化不良、胃酸过多。 【用法】：每次1～3片，3 次/日 【注意事项】：①不宜与胃蛋白酶合剂、维生素等酸性药物合用。②密闭阴暗贮藏，否则逐渐变质，一部分碳酸氢钠变为碳酸钠。 【不良反应】：口服后可能因产生大量二氧化碳而使胃扩张，并刺激溃疡。 第二节H2受体阻断剂：可拮抗组胺引起的胃酸分泌

详解HTML5中rel属性的prefetch预加载功能使用

这篇文章主要介绍了HTML5中rel属性的prefetch预加载功能使用,特别是在用户第一次访问Web页面浏览器尚无缓存的时候,prefetch可以用作加速,需要的朋友可以参考下在HTML5中,有个很有用但常被忽略的特性,就是预先加载(prefetch),它的原理是: 利用浏览器的空闲时间去先下载用户指定需要的内容,然后缓存起来,这样用户下次加载时,就直接从缓存中取出来,效率就快了. 举个例子说明:比如要预先加载某个页面,可以这样: XML/HTML Code <link rel="prefetch" href="https://www.wendangku.net/doc/4e14200293.html,/"> <!-- Firefox --> 但如果是google的话,要用另外的一个名称,即: XML/HTML Code <link rel="prerender" href="https://www.wendangku.net/doc/4e14200293.html,/"> <!-- Chrome --> 即使在不支持的浏览器,用了这个特性其实是不会出错的,只不过浏览器解析不到而已, 所以,如果你感觉能有办法预先预测到用户期望点的页面(比如用户看最新的受欢迎的热图,他可能看了第一页后,会继续看下一页,这个时候就可以用预先加载这个特性了).比如XML/HTML Code <link rel="prefetch" href="<?php echo get_next_posts_page_link(); ?>"> 而单独取一张图片也是可以的,比如: XML/HTML Code <link rel="prefetch" href="/images/test.jpg"/> 有了浏览器缓存，为何还需要预加载？ 1.用户可能是第一次访问网站，此时还无缓存 2.用户可能清空了缓存 3.缓存可能已经过期，资源将重新加载 4.用户访问的缓存文件可能不是最新的，需要重新加载 5.Chrome 的预加载技术 现在的chrome 聪明到根据你的浏览记录，预测到你可能访问或搜索哪些网站，在你打开网站之前就加载好了一些资源了。 举个栗子，当你在搜索框输入"amaz" 时，它猜测到你可能要访问https://www.wendangku.net/doc/4e14200293.html,，可能就帮你加载了这个网站的一些资源。 如果这个预测算法精准的话，就能大大地提高用户的浏览体验了。 DNS prefetch 我们知道，当我们访问一个网站如https://www.wendangku.net/doc/4e14200293.html, 时，需要将这个域名先转化为对应的IP 地址，这是一个非常耗时的过程。 DNS prefetch 分析这个页面需要的资源所在的域名，浏览器空闲时提前将这些域名转化为IP 地址，真正请求资源时就避免了上述这个过程的时间。 XML/HTML Code <meta http-equiv='x-dns-prefetch-control' content='on'> <link rel='dns-prefetch' href='https://www.wendangku.net/doc/4e14200293.html,'> <link rel='dns-prefetch' href='https://www.wendangku.net/doc/4e14200293.html,'> <link rel='dns-prefetch' href='https://www.wendangku.net/doc/4e14200293.html,'> <link rel='dns-prefetch' href='https://www.wendangku.net/doc/4e14200293.html,'> <link rel='dns-prefetch' href='https://www.wendangku.net/doc/4e14200293.html,'>

ES系列以太网交换机使用说明(Content)

第一章产品介绍 1.1产品简介 ES系列快速以太网交换机是款完全符合IEEE 802.3 Ethernet 标准，并且满足工业生产的苛刻要求的高性能交换机，它为建立小型、中型、大型网络尤其是工业自动化控制网络、小区社区网络接入提供了最具性价比的组网解决方案。本系列交换机目前包括ES-24/ES-24F和ES-08三款交换机,其中ES-24F提供光模块接口扩展。 在本系列交换机中，所有的端口都支持自适应功能，与任何10Mbps 或100Mbps ，全双工或半双工的以太网设备相连都能保证正常工作，并可独享速率，大幅提升网络性能。采用最新的“自动交叉线（Auto-Cross-Over）技术，能自动检测双绞线为直通线或交叉线，任何线与任何口都可以相连，所有端口都可以作级联口。本系列交换机还可以扩展1 或2 口100BASE-FX SC/ST 光纤模块，用来连接远距离的交换机或服务器，最长可延伸2公里(多模)或20公里以上(单模)距离，其独立的模块口不占用其它端口。 1.2 装箱清单 先检查包装是否完全如下列附件，如果任一附件遗失或受损，请与您的经销商联系并保留原包装，包装中有以下附件： ·一台以太网交换机 · L型固定架两个 ·镙钉六枚 ·黏性胶垫四个 ·使用手册 1.3 产品特性和规格 产品特性 ● 符合IEEE 802.3 标准 ● 流控方式：全双工采用IEEE 802.3x 标准，半双工采用Backpressure标准 ● 存储-转发体系结构 ● 具有8/24 个10Base-T/100Base-TX RJ-45 端口(支持MDI/MDIX 自动翻转功能) ● 提供2个扩展插槽，支持100M光纤/UTP模块卡和宽带路由模块卡 ● 背板带宽大于4.8G ● 转发速率：10M 14，880pps 100M 148,800pps ● 支持4K MAC地址空间 ● 缓冲区容量6M ● 每一端口支持地址学习功能，并允许设置动态地址老化时间 ● 支持静态MAC地址表的管理及静态MAC地址绑定功能 ● 能提供端口安全控制、端口监控等设置功能 ●提供多种电源支持，包括AC 220V，DC 220V和DC 110V ●默认电源支持AC 220V/DC220V自适应 ●在-25 oC至70 oC间可保证正常工作 ●在温度为4 0 oC，湿度为95％的湿热环境（无凝结）下可保证工作正常 ●可在10V/m的强磁场辐射环境下正常工作 ●6Kv接触放电（静电干扰）下工作无影响

正确使用生活中常用药物

正确使用生活中常用药物 姓名：11级公管2班吴山 摘要：掌握常用口服药品的服用时间及原因。掌握常用口服药品的正确使用方法和注意事项。掌握外用剂型的正确使用方法和注意事项。掌握常用外用药品的正确使用方法和注意事项。掌握常用注射剂的正确使用方法和事项。 关键词：常用药物，药物的使用方法和注意事项。 正文：目前，我国医药行业朝着快速、高投入、不断更新换代的方向发展。特别是随着各种新药、仿制药不断涌现，在充分满足人们防病、治病需求的同时，也导致药物不良反应问题日益严重。为此，如何保障大众合理正确使用药物已成为一个亟待解决的重点问题。执业药师应该在合理正确使用药物方面起指导作用。正确合理使用药物必须对药物的特点、性质、适应性、用法用量、用药时间、注意事项等等有完全的了解和掌握。 合理正确使用药物首先要了解、掌握国家基本药物。早在1979年，我国就已开始实行基本药物政策。只有了解、掌握基本常用药物，进一步学习基本常用药物，才能更有效地保障大众合理用药。本文叙述了部分国家基本药物的合理正确使用方法，主要对常用口服药物、常用外用药物、常用注射的药物进行阐述。 一、基本药物的概念 基本药物的概念是世界卫生组织(WHO)于1977年提出的，其英文名称为Essential Medicine。2002年1月，WHO将基本药物定义为：能优先满足人们卫生保健需求的药物，是按照一定的遴选原则，经过认真筛选确定的、数量有限的药物。我国对基本药物的遴选原则是“临床必需、安全有效、价格合理、应用方便、中两药并重”。1979年，我国政府制订了《国家基本药物目录》，迄今为

止已修订了4版。除此之外，还制订了《国家基本药物临床手册》及《国家基本药物中药制剂临床指南》。

Content

Content [点拨] content adj.满足的，满意的 常用短语： ① be content with = be satisfied with = be pleased with对……满意 ② be content to do sth. = be willing to do sth. = be ready to do sth. 乐意做某事 例句： Are you content / satisfied / pleased with your work? 你对你的工作满意吗？ The old man is content / willing / ready to live in the countryside. 那位老人乐意住在乡村。 n. ①满足（不可数） 常用短语：to one’s heart’s content 尽情地 例句： The child is singing to his heart’s content. 这个孩子尽情地唱歌。 ②（书、文章、节目、演说等的）内容（常用单数形式）；容纳的东西（常用复数形式） 例句： I like the style of his wr iting but I don’t like the content. 我喜欢他的写作风格，但是不喜欢他写的内容。 He looked at the contents of the bag. 他看了看袋子里的东西。 ③目录（常用复数形式） 例句： The contents of a book can help you understand the book. 书的目录可以帮助你了解一本书。 ④含量, 容量（不可数） 例句：

http中content-type头值-

http中content-type头值－(MIME类型) 常见文件http中content-type头值－(MIME类型) .ppt–application/mspowerpoint .ai–application/postscript .aif–audio/x-aiff .aifc–audio/x-aiff .aiff–audio/x-aiff .asc–text/plain .au–audio/basic .avi–video/x-msvideo .bcpio–application/x-bcpio .bin–application/octet-stream .c–text/plain .cc–text/plain .ccad–application/clariscad .cdf–application/x-netcdf .class–application/octet-stream .cpio–application/x-cpio .cpt–application/mac-compactpro .csh–application/x-csh .css–text/css .dcr–application/x-director .dir–application/x-director .dms–application/octet-stream .doc–application/msword .drw–application/drafting .dvi–application/x-dvi .dwg–application/acad .dxf–application/dxf .dxr–application/x-director .eps–application/postscript .etx–text/x-setext .exe–application/octet-stream .ez–application/andrew-inset .f–text/plain .f90–text/plain .fli–video/x-fli .gif–image/gif .gtar–application/x-gtar .gz–application/x-gzip .h–text/plain .hdf–application/x-hdf .hh–text/plain

CAD直接修改文字或块的属性内容程序

(defun c:cht(/ e ent en newt oldt ent1) (setq e (car (entsel "\nPick a text or a attrib: "))) (if (/= e nil) (progn (setq ent (entget e)) (cond ((and (= (cdr (assoc 0 ent)) "INSERT") (= (cdr (assoc 66 ent)) 1)) (progn (setq en (entget (setq ent (entnext e)))) (setq oldt (cdr (assoc 1 en))) (setq newt (getstring T (strcat "\nNew text <" oldt ">:"))) (if (= newt "") (setq newt oldt)) (setq ent1 (subst (cons (car (assoc 1 en)) newt) (assoc 1 en) en)) (entmod ent1) (entupd ent) )) ((= (cdr (assoc 0 ent)) "TEXT") (progn (setq oldt (cdr (assoc 1 ent))) (setq newt (getstring T (strcat "\nNew text <" oldt ">:"))) (if (= newt "") (setq newt oldt)) (setq ent1 (subst (cons (car (assoc 1 ent)) newt) (assoc 1 ent) ent)) (entmod ent1) )) (T (princ "\nError: Not a text or not a block or no attrib in block !")) ) ) ) (princ) )

文件属性详解

linux中各种文件类型 普通文件（- regular file） (1)文本文件。文件中的内容是由文本构成的，文本指的是ASCII码字符。文件里的内容本质上都是数字（不管什么文件内容本质上都是数字，因为计算机中本身就只有1和0），而文本文件中的数字本身应该被理解为这个数字对应的ASCII码。常见的.c 文件, .h文件 .txt文件等都是文本文件。文本文件的好处就是可以被人轻松读懂和编写。所以说文本文件天生就是为人类发明的。 (2)二进制文件。二进制文件中存储的本质上也是数字，只不过这些数字并不是文字的编码数字，而是就是真正的数字。常见的可执行程序文件（gcc编译生成的a.out，arm-linux-gcc编译连接生成的.bin）都是二进制文件。 (3)对比：从本质上来看（就是刨除文件属性和内容的理解）文本文件和二进制文件并没有任何区别。都是一个文件里面存放了数字。区别是理解方式不同，如果把这些数字就当作数字处理则就是二进制文件，如果把这些数字按照某种编码格式去解码成文本字符，则就是文本文件。 (4)我们如何知道一个文件是文件文件还是二进制文件？在linux系统层面是不区分这两个的（譬如之前学过的open、read、write等方法操作文件文件和二进制文件时一点区别都没有），所以我们无法从文件本身准确知道文件属于哪种，我们只能本来就知道这个文件的类型然后用这种类型的用法去用他。有时候会用一些后缀名来人为的标记文件的类型。 (5)使用文本文件时，常规用法就是用文本文件编辑器去打开它、编辑它。常见的文本文件编辑器如vim、gedit、notepad++、SourceInsight等，我们用这些文本文件编辑器去打开文件的时候，编辑器会read读出文件二进制数字内容，然后按照编码格式去解码将其还原成文字展现给我们。如果用文本文件编辑器去打开一个二进制文件会如何？这时候编辑器就以为这个二进制文件还是文本文件然后试图去将其解码成文字，但是解码过程很多数字并不对应有意义的文字所以成了乱码。 (6)反过来用二进制阅读工具去读取文本文件会怎么样？得出的就是文本文字所对应的二进制的编码。 目录文件（d directory） (1)目录就是文件夹，文件夹在linux中也是一种文件，不过是特殊文件。用vi打开一个文件夹就能看到，文件夹其实也是一种特殊文件，里面存的内容包括这个文件的路径，还有文件夹里面的文件列表。 (2)但是文件夹这种文件比较特殊，本身并不适合用普通的方式来读写。linux中是使用特殊的一些API来专门读写文件夹的。 字符设备文件（c character） 块设备文件（b block） (1)设备文件对应的是硬件设备，也就是说这个文件虽然在文件系统中存在，但是并不是真正存在于硬盘上的一个文件，而是文件系统虚拟制造出来的（叫虚拟文件系统，如/dev /sys /proc等） (2)虚拟文件系统中的文件大多数不能或者说不用直接读写的，而是用一些特殊的API产生或者使用的，具体在驱动阶段会详解。 管道文件（p pipe） 套接字文件（s socket） 符号链接文件（l link）

常用药品名称、用法及用途

常用药品的名称用法及用途 1.盐酸肾上腺素(负肾，AD，1mg/1ml) 1）作用：兴奋α、β两种受体，使心肌收缩力增强，心率加快，升高血压，心肌耗氧量增加，松弛支气管平滑肌，解除支气管痉挛。2）适用于：过敏性休克，心脏骤停，荨麻疹，支气管哮喘、粘膜或齿龈的局部止血等。 2.酒石酸去甲肾上腺素（正肾，NA，2mg/1ml） 1）作用：显著地增强心肌收缩力，使心率增快，心输出量增多；使除冠状动脉以外的小动脉强烈收缩，引起外周阻力明显增大而使血管收缩，升高血压。2）适用于：急性心肌梗塞，体外循环、嗜铬细胞瘤切除等引起的低血压。 3.硫酸异丙肾上腺素（喘息定，SOprenaline,1mg/2ml） 1）作用：兴奋心脏，改善心脏传导，增加回心血量，升高血压，使脉压增大，扩张内脏血管，扩张支气管平滑肌。2）适用于：缓慢性心律失常、支气管哮喘、中毒性休克及心脏房室传导阻滞。 4.尼可刹米（可拉明，Nikethamide，0.375g/1.5ml） 1）作用：兴奋延髓呼吸中枢，使呼吸加深加快。2）适用于：中枢性呼吸衰竭，继发性呼吸抑制及循环衰竭。 5.山梗菜碱（洛贝林，Lobeline，3mg/1ml） 1）作用：刺激颈动脉窦和主动脉体化学感受器，反射地兴奋呼吸中枢，使呼吸加深加快。2）适用于：新生儿窒息、CO引起的窒息以及肺炎等引起的呼衰。 6.去乙酰毛花甙（西地兰，Deslanoside,0.4mg/2ml） 1）作用：增强心肌收缩力，减慢心率与传导，正性肌力，利尿。2）适用于：急性充血性心力衰竭，心房颤动、扑动，阵发性室上性心动过速。 7.多巴胺（Dopamine,20mg/2ml） 1) 作用：增加心排血量，加快心率，收缩外周血管，扩张内脏血管。2）适用于：各种休克的治疗，对伴有肾功能不全、心排血量降低，周围血管阻力增高而已补充血容量的更有意义。 8.阿托品（Atropine,1mg/1ml） 1)作用：解除平滑肌痉挛，抑制腺体分泌，散大瞳孔，升高眼压；解除血管痉挛，改善微循环而起到抗休克的作用，并能兴奋呼吸中枢。2）适用于：内脏绞痛、早搏、感染性休克、急性微循环障碍、严重心动过缓，有机磷农药中毒、麻醉时抑制腺体分泌、阿—斯综合征。 9.间羟胺（阿拉明，Metaraminol,10mg/1ml） 1)作用：兴奋α受体，缓慢持久地收缩血管和中度增强心肌收缩力。2）适用于：各种休克及手术时低血压、心梗性休克。 10.硝酸甘油（Nitroglycerine,5mg/1ml） 1）作用：扩张静脉和小动脉，减少回心血量，降低心脏前后负荷，减少心肌耗氧，改善冠状动脉供血；松弛血管平滑肌，扩张动静脉血管，缓解心绞痛，降低血压。2）适用于：治疗肺水肿，指端静脉痉挛及预防心绞痛。 11.普罗帕酮（心律平，Propafenone,35mg/10ml） 1)作用：抗心律失常，松弛冠状动脉及支气管平滑肌局麻作用；增加冠脉血流及轻中度抑制心肌收缩力作用。2）适用于：室早、阵发性室速及预激综合征。 12.呋塞米（速尿，Furosemide,20mg/2ml） 1)作用：抑制髓袢升支的髓质部对钠、氯的重吸收，促进钠、氯、钾的排泄和影响肾髓

CSS和DOM属性用法速查手册

CSS和DOM属性用法速查手册 CSS属性用法速查手册 -------------------------------------------------------------------------------- !important增加特定规则的重要性。 :active设置当链接处于激活状态时a元素的样式。 :first-letter在对象的第一个字符上应用一个或多个样式。 :first-line在对象的第一行上应用一个或多个样式。 :hover设置当用户将鼠标指针悬停在链接上时a元素的样式。 :link设置当链接最近没有访问过时a元素的样式。 :visited设置当链接最近访问过时a元素的样式。 @charset设置外部样式表的字符集。 @font-face设置要嵌入HTML文档的字体。 @import导入一个外部样式表。 @media设置styleSheet对象中一组规则的媒体类型。 @page设置styleSheet中页面框的尺寸、方向和边距。 abbr设置或获取对象的缩写文本。 accelerator设置或获取表明对象是否包含快捷键的字符串。 accept设置或获取以逗号分隔的内容类型列表。 acceptCharset设置或获取处理表单的服务器必须接受的输入数据所用的字符编码方式列表。 accessKey设置或获取对象的快捷键。 action设置或获取表单内容要发送处理的URL。 activeElement获取当父document拥有焦点时获得焦点的对象。 additive设置或获取表明动画是否附加到其它动画的值。 align设置或获取对象针对其邻接文本如何排列。 align设置或获取标题或标志的排列。 align设置或获取表格排列。 align设置或获取对象相对于显示或表格的排列方式。 aLink设置或获取元素中所有激活链接的颜色。 alinkColor设置或获取元素中所有激活链接的颜色。 allowTransparency设置或获取对象是否可为透明。 alt设置或获取用于替代图像的文本。 altHTML设置可选的若对象装载失败时要执行的替换HTML脚本。 altKey设置或获取Alt键的状态。 altLeft设置或获取左Alt键的状态。 appCodeName获取浏览器的代码名称。 APPLICATION表明对象的内容是否为HTML应用程序(HTA)，从而免除浏览器的安全模型。 appMinorVersion获取应用程序的次版本值。 appName获取浏览器的名称。 appVersion获取浏览器运行的平台和版本。

WEB前端程序猿必看的meta标签汇总

WEB前端程序猿必看的meta标签汇总 对于一般的前端工程师来说，知道一些常用的就好了，但是突然有一天，BOSS提出一些比较另类的需求，譬如：忽略页面中的数字识别为电话，忽略email识别这一类的需求等，我们又不得不去满足Ta。那些写过的代码一股脑的上涌，但是这并没有用，瞬间大脑一片空白，对这个需求一点印象。这时，如果有一位大牛能点拨一下，是不是会有一种想以身相许的冲动。呵呵……这时候小编的这编汇总就可以帮到您。 小编查阅大量国内外相关资源，终于总结出了以下的文章，希望以下的内容对您有帮助。那么我们就直奔我们的主题吧！来，上代码。 首先，我们先来了解一下，什么是meta标签？ 元素可提供有关页面的元信息（meta-information），比如针对搜索引擎和更新频度的描述和关键词。 以上是w3c上的解释。元信息，是用来描述数据的数据，也就是用来描述当前页面的一些信息。例如：定义页面的一些描述信息、文件编码、关键词、作者等等。 一、标签包含的属性 https://www.wendangku.net/doc/4e14200293.html,属性 name 属性提供了名称/值对中的名称（而后面要说的content属性则是该名称对应的值）。假如我们把页面想象成一个我们生活中的实物的话，拿个人简历来作类比，诸如姓名、性别、籍贯、技能、项目经验等等这些都可以看做是个人简历的name属性，content属性相当于这些名称对应的值。 基本语法结构： 2.content属性 content 属性提供了名称/值对中的值。该值可以是任何有效的字符串。 content 属性始终要和name 属性或http-equiv 属性一起使用。 3.http-equiv属性 equiv是equivalent的简写，是相等的，等价物的意思。不难理解，相当于http的文件头作用，它可以向浏览器传回一些有用的信息，以帮助正确和精确地显示网页内容，与之对应的属性值为content，content中的内容其实就是各个参数的变量值。（http不了解的同学，可以找找相关的资料了解一下）。 基本语法结构： 二、标签常用的名称/值对 1.keywords(关键字) 说明：用于告诉搜索引擎，你网页的关键字。

临床常见药物用法

盐酸多巴胺注射液【20mg 2ml/支】 【用法】1-5μg/kg*min，每15-30min增加1-4μg/kg*min 【泵入】kg×3+NS 至50ml，1ml/h=1μg/kg*min 【滴入】5%GS 70ml 多巴胺 300mg ，1.2ml/h=1μg/kg*min 【中日急诊】5%GS 100ml 多巴胺 300mg ，5ml/h起(约11.5mg/h,对60kg约3.2ug/kg/min) 盐酸乌拉地尔注射液【亚宁定，25mg 5ml/支】 【用法】25mg+10mlNS慢推一半，15分钟后再推另一半，然后100-400μg/ min(6-24mg/h)维持 【泵入】乌拉地尔100mg NS 30ml ， 3ml/h=6mg/h 【滴入】乌拉地尔 50mg NS 250ml ，10滴/min=30ml/h=6mg/h 【中日急诊】NS 100ml 乌拉地尔 200mg,5ml/h起（约7mg/h） 注射用生长抑素【思他宁3000ug/支*】 【用法】上消化道出血：250μg缓慢注射（>3min），止血后250μg/h维持3-4天，但<120h。 急性胰腺炎：250μg/h维持5-7天 【泵入】生长抑素 6mg NS48ml ，2ml/h=240μg/h；先入2ml。 【滴入】NS或GS 500ml 生长抑素 3mg，ivgtt连续静滴12h。 奥曲肽注射液【善宁，0.1mg 1ml/支】 【用法】25μg缓慢注射，25-50μg/h维持3-4天 【泵入】奥曲肽 0.6 NS 48ml ，2ml/h=24μg/h；先入2ml。 【皮下】预防胰腺手术后并发症，0.1mg 皮下 Q8h×7天，第一次用药至少在术前1小时进行。 注射用甲磺酸加贝酯【100mg/支，70.39元】 【滴入】急性轻型胰腺炎或重症辅助： 加贝酯 100mg 5%GS或林格500ml ，ivgtt（<1mg/kg/h） tid×3天，改为100mg/日，共6-10天 注射用乌司他丁【天普洛安，10万U/支，134.99】 急性胰腺炎、慢性复发性胰腺炎的急性恶化期： 【滴入】5%GS或0.9%NS 500ml 乌司他丁 10万U ，ivgtt 1-2h入 Qd-Tid，随症状改善减量 急性循环衰竭： 【滴入】 5%GS或0.9%NS 500ml 乌司他丁 10万U ，ivgtt 1-2h入 Qd-Tid 【静推】2ml 0.9% NS 乌司他丁 10万U ，缓慢静脉推注 Qd-Tid

高中英语语法知识之It的用法总结教学内容

高考英语语法知识之It 的用法总结 1．It is + 被强调部分+ that ... 该句型是强调句型。将被强调的部分放在前面，其它部分置于that之后。被强调部分可以是主语，宾语，状语。强调的主语如果是人，that可以由who换用。 如果把这种句型结构划掉后，应该是一个完整无缺的句子。这也是判断强调句型与其它从句的方法。 It was they that (who) cleaned the classroom yesterday. It was in the street that I met her father. 2．It was not until + 被强调部分+ that ... 该句型也是强调句型。主要用于强凋时间状语，译成汉语＂直到．．．才．．．＂，可以说是not ... until ... 的强调形式。 It was not until she took off her dark glasses that I realized she was a famous film star. = Not until she took off her dark glasses did I realize she was a famous film star. = I didn’t realize she was a famous film star until she took off her dark glasses. 3．It is clear ( obvious, true, possible,certain....) that ..... 该句型中it 是形式主语，真正的主语是that 引导的主语从句，常译为＂清楚（显然，真的，肯定．．．）＂是主语从句最常见的一种结构。 It is very clear that he’s round and tall like a tree.= That he’s round and tall like a tree is very clear. 4. It is important ( necessary, right, strange, natural...) that ... 由于主句中的形容词不同，that 后的从句中要用虚拟语气（should + 动词原形），should 可以省去。 It is important that we (should) learn English well. It is necessary that he (should) remember these words. 5. It is said (reported, learned....) that ... 该句型中的it 仍是形式主语，真正主语是that 引导的主语从句。该结构常译为＂据说（据报道，据悉．．．）＂。It is said that he has come to Beijing. 6. It is suggested ( ordered ... ) that ... 主句中的过去分词是表示请求，建议，命令等词时，that后的从句要用虚拟语气（should + 动词原形），should 可以省。常译为＂据建议；有命令．． It is suggested that the meeting ( should ) be put off. It was ordered that we ( should ) arrive there in two hours. 7. It is a pity ( a shame ... ) that ... 该句型中，that后的从句一般用虚拟语气（should + 动词原形），should可省去．表示出乎意料，常译为＂竟然＂。没有这种意义时，则不用虚拟语气。 It is a pity that such a thing ( should ) happen in your class. It is a pity that he is ill. 8. It is time ( about time ,high time ) that ... 该句型中that 后的从句应该用虚拟语气，值得注意的是①常用过去时态表示虚拟．②有时也用should + 动词原形，should 不能省。常译为＂是（正是）．．．的时侯．．．＂。 It is time that children should go to bed. = It is time that children went to bed. 9. It is the first ( second ... ) time that ... 该句型中的that 从句不用虚拟语气，而用完成时态。至于用什么完成时态，由主句的谓语动词的时态决定。如果是一般现在时，后面从句用现在完成时态；如果是一般过去时，后面从句则用过去完成时态。该结构中that 可以省去；it有时用this / that 替换．常译为＂这是某人第几次做某事了＂。 It is the first time I have been here. = This is the first time I have been here. 10 It is the +形容词最高级+ 名词+ that + ….. 该句型中的that 从句不用虚拟语气，而用完成时态。至于用什么完成时态，由主句的谓语动词的时态决定。如果是一般现在时，后面从句用现在完成时态；如果是一般过去时，后面从句则用过去完成时态。该结构中that 可以省去；it有时用this / that 替换．常译为＂这是某人做过的最…的事情＂。