Recently, I have read a paper Automated reconstruction of 3D scenes from sequences of images. The paper is mainly about 3D reconstruction. In this paper, a system is presented which automatically extracts a textured 3D surface model from a sequence of images of a scene. The key techniques consist of three important parts: projective reconstruction, self-calibration and dense depth estimation. The model can be built by these techniques. At last, the paper gives some examples which using the method the paper presented and some photos are given to testify it.
The outline of the passage is as follows:
Firstly, the author introduces the new research result in the abstract which describe the main idea of the paper. Secondly, in the introduction, the author describes the backgrounds of the development and gives some instruction of the whole system. Thirdly, the overview of the method that the paper is mainly about is given. Then, the author tells us the details about the three important parts which is the most crucial part of the paper. Then the author tells us how to make good use of the method to build the model and gives some example. Based on this, conclusion and further research is presented to us. At last of the paper, acknowledgements and references is listed.
The important sentence patterns are as follows:
1.Typically, it is assumed that the same camera is used throughout the sequence and that the intrinsic camera parameters are constant.
2.Since a projective calibration is not sufficient for most applications, researchers tried to find ways to automatically upgrade projective calibrations .
3 A metric reconstruction calibration is obtained by transforming the projective reconstruction to a solution for which all the constraints on the camera intrinsic parameters are satisfied.
4 Once the camera projection matrix has been determined, the reconstruction is updated.
5 A possible solution consists of not only relating the actual view with the previous one, but also with other close views.
6 The pairwise disparity estimation allows to compute image-to-image correspondence between adjacent rectified image pairs, and independent depth estimates for each camera viewpoint.
7 The 3D surface acquisition technique that was presented in the previous sections, can readily be applied to archaeological sites.
The problem is that this kind of overview model is too coarse for use in realistic walk-throughs around the site or for looking at specific monuments.
8 Due to the ease of acquisition and the obtained level of detail, one could reconstruct every building block separately.
The important English words and corresponding Chinese are as follows:
calibration 标定
extract 提取
projective reconstruction 映射关系重建
dense depth estimation 稠密深度匹配
dynamic programming 动态规
subsequent 后来的
epipolar 对极
rectification 修正
consecutive 连续的
intensity 剧烈变化
aligned 排列的
optimal 最优的
updated 校正过的
conic 二次曲线
arbitrary rentide任意的
polar-ordinates 极坐标
homograph 单应的
ambiguity 模糊的
bidirectional 双向的
uniqueness 唯一的
cross-correlation 互相关
pairwise disparity estimation 成对视差估计
occlusion 闭塞
segmentation 分割
morphological filtering 形态滤波
spurious 假的
mesh 网格
second-order spline 二次样条
texture mapping 纹理映射
virtualise 虚拟化
vertical resolution 垂直分解度
feasibility可行性
The passage’s writing characteristic is as fallows:
As the passage is an academic writing, the passage is very precise. In academic writing, the author usually applies phrases, composite words, long sentences and initialisms, the tendency to avoid the use of first and second pronouns, the avoidance of contractions, greater conciseness through the avoidance of repetition and the use of noun phrases to replace subordinate clauses, the use of more tentative forms of expression and the use of more precise, formal vocabulary for the passage.
Important complements
deal with 处理, 研究, 讨论
it is assumed that……
A possible solution consists of not only relating the actual view with the previous one, but also with other close views.
We would like to thank……
Due to the ease of acquisition and the obtained level of detail, one could reconstruct every building block separately.
In the case of Sagalassos, some building hypothesis were translated to CAD .
Since these reconstructions are generated automatically and the on-site acquisition time is very short, several new applications come to mind.
apply to 适用于,应用于
product “visits”.
define as 定义为
concentrate on 集中
draw one’s conclusion 得出结论
consist in 包含
The extension of the paragraph:
The paper is so long that we should extract the main thread that we can read it easier.
As usual, the paper begins with a abstract which is useful for us to know what is the paper talking about. Then there is a introduction about the background of why the method the paper is about was given. Then is the center of the paper, it mainly presents the author’s work on the problem. And at last it is the result and some examples about the realities of the method. This the often used model of academic paper.
After reading the academic passage, I learn a lot of knowledge from it. Firstly, I know how to write an English academic paper. Secondly, I get knowledge about my major. What’s more, we must take care of the vocabulary which is used in academic writing to express the our methods.