The method of 3D reconstruction in civil construction involves the procedure of development of Three - dimensional models from three numbers of two - dimensional orthographic views.

3D reconstruction method in building construction technology faces certain advantages as well as limitations when applied. This building construction technology in the aspect of application is unique in nature.

This method makes use of knowledge and their combination which is merely based on the drawing interpretation. This is done such a way so that the recognition and the reconstruction of the engineering components are possible from the description given in simplified form in the respective engineering drawings.

Contents:

**3D Reconstruction Technique in Construction**

Itâ€™s a common practice to make use of two - dimensional drawing to describe the three - dimensional construction component mainly in the construction industry. This is still a practice to bring the 2D drawings, even though there are CAD software packages that will let the user to draw the 3D components. A 3D component is easily understood by drawing its respective 2D drawing or its sectional drawing in some situations.

As mentioned above an idea about the 3D components is obtained by the 2D orthographic views. This can be clearly explained with the help of an example:

**Example 1:**

The quantity surveyor has to reconstruct the 3D models mentally so that he doesnâ€™t miss out any relevant quantities and carry out quantity surveying. This will let him measure the relevant quantities without missing out any work elements. Taking these quantities is very tedious process and are prone to errors.

So, the development of a computer program to measure the quantities must have the ability to reconstruct the 2D orthographic views to corresponding 3D models automatically and then carry out the measurement.

**History of 3D Reconstruction Technique**

The research on 3D reconstruction almost started in 1970s. Thereafter continuous research was carried out to bring to a proper solution. Among many of the algorithms that were developed, most of them could only transform the 2D view into three dimensional planar objects only.

The reconstruction of 3D elements of curve and improper shapes were very difficult to generate through the 2D orthographic views.

**3D Reconstruction Algorithm Development**

As the aspect of developing algorithm vary with the aspects considered, there are certain major tasks that should be adopted to generate the 3D model from the respective 2D orthographic views:

**Step 1:** The initial procedure is to transform the two - dimensional points into respective three - dimensional vertices.

**Step 2:** The second procedure is to generate the edges of the 3D element from the 3D vertices that were developed in the step 1.

**Step 3:** The next step once the 3D vertices and the 3D edges are obtained is the development of faces and its construction. This is developed from the 3D edges.

**Step 4:** From the 3D faces developed and constructed, the 3D polyhedron is formed.

A non-existent or a false vertices, edges, and faces can be developed in the step 3. So, the identification of such false elements and their removal so that the final 3D model is true is one of the challenging activity in the development of 3D reconstruction algorithm.

Another difficulty faced in the development of 3D reconstruction algorithm is the characteristics of the engineering or the construction drawings. There are many components present in the construction drawings including the steel reinforcement bars that is very difficult to be drawn in their true dimensions in 2D figure. So, in 2D, the designers makes use of simplified figures so that they can be represented easily. This can be clearly explained with the help of following example.

**Example 2:**

A single line representation is made for reinforcement bars that are attached with a text telling the number and the size of the respective bar. These simplifications are very difficult to be reconstructed into 3D model from their 2D orthographic views.

A study conducted by Longguet-Higgins in 1981 showed that the if the two images of a particular scene containing 8 points that is located in both the projections, then the respective structure can be computed from this 8 sets of corresponding image coordinates.

This is one of the study that laid the future to 3D reconstruction. The 3D reconstruction in construction industry is a very new topic that is on its way of gaining popularity.

**3D Reconstruction Process**

The below flowchart shows the 3D reconstruction Process in a general format.

**Fig.1: Flowchart Showing the General 3D Reconstruction Process**

**Process 1: Image Acquisition**

The steps except the first one is used or mentioned only when a 3D reconstruction process is employed. The initial step can also be called as the input step.

This step involves the capturing of the images with the help of cameras. The camera can be either calibrated or non-calibrated based on the applications. Most of research studies and case study conducted make use of cameras that are calibrated.

**Process 2: ****Extraction of Features and ****Establish Correspondences**

The recognition of features involves the determining the key points in a particular image. The extraction is performed by certain algorithms called the **feature detector**. These are evaluated based on their repetitive nature and the information content.

Here, repeatability means the ability of the feature detector to determine the same point in different images.

The information content means the ability of the detector to distinctly detect the points. A good performing feature detector must have higher levels of both the repeatability and the information content properties.

**Example 3:**

One such detector was Moravecâ€™s Corner Detection algorithm that help a mobile robot to cut the corners of indoor and the outdoor spaces.

Once the feature correspondences are obtain, the epipolar geometry is used to find the points in the 3D and the computation of 3D structure is made.

**Conclusion -3D Reconstruction Technique **

The 3D reconstruction technique is an example on how reliable computer vision based algorithms are employed for solving the real - life problem. The 3D reconstruction techniques are gaining their application in Quantity analysis as well as disaster analysis of civil structures.

FIT3D and Autodesk 123D Catch are potential useful 3D reconstruction toolkits. Better results were obtained by researchers from the AUTODESK 123D compared with FIT3D.

The Autodesk 123D have self â€“ calibration capabilities. This program also have user friendly tools in order to edit or view the 3D model. The task is made easy by the 3D reconstruction as it requires laptop or computer, a software for model development and a digital camera.

Following are the requirements that must be attained in order to undergo best reconstruction task:

- Choose the Proper methodology
- Plan the locations whose photos have to be taken before the overlapping of the images is performed.
- If any obstruction covers an essential feature of the structure, a reference object has to be placed.