The system utilizes a technology call Augmented Reality (AR) which uses a mobile device’s camera to superimpose a computer-generated image or 3D model on to a user’s view of the real world. The traditional Augmented Reality tools which were used until now relied upon a pre-defied marker on to which the computer-generated object can be placed. The system is utilizing the marker-less AR technology. This technology analyzes the user’s surrounding and define track points which can then be used to lock 3D models into place. This method will provide user with greater mobility and allow him/her to view and customize the placed object in any direction and distance without always having to focus on a marker.
This application is research four difference areas.
Better Augmented Reality Experiance
Sri Lanka is lacking behind providing a better customer satisfactory solution. Several researches have been carried out in fields related to similar problem but no practical solution was proposed as of yet. When considering Augmented Reality, solution propose were based on predefined markers (A image or pattern of the real world pre-stored in a database) when compared and detected in the real world on which augmented 3D objects are displayed. The current solutions which are offered for marker-less AR lack the reliability and support necessary to be use in mobile applications.
find a suitable furniture that matches with the surrounding
finding a suitable furniture that matches with the surrounding of the area that someone supposed to keep the furniture is problematic. It may be because of non-matching colors, furniture shape or because of the space limitations in furniture placement. Among those issues the color of the furniture plays a prominent role where the issues can be minimized by selecting a matching furniture at first place.
Identify Real word Scale
we use mobile camera for capturing scenes taking dimensions of each object that present in the scene is bit of troublesome because taking those dimensions depend on many factors like focal length of camera height of the camera view etc. Deriving dimensions like height width and length form a camera feedback is another key point of this application.
Voice Recogntion and Augmented Audio
Today, society is more familiar with voice command and voice recognition systems. When it comes augmented reality field it is more suitable to use voice command. In Android Environment does not provide a direct means to recognize voice inputs. Therefore the device cannot recognize the speech to overcome this we need to use a different approach.
“Marker-less Augmented Reality Interactive Furniture layouting” will be offering solution taking into account all the prior mentions research gap to over-come the issues customers face when purchasing furniture.
Designing furniture according to the exact requirements of the customer has always been a pending problem in the furniture manufacturing industry. Customers who also request ready-made furniture will not know if the selected furniture actually suit them till it has been bought and placed in their home. If they are not satisfied with the furniture they will either refund it or if it cannot be refunded, will have to make do with it. This results customer dissatisfaction as well as the decline of marketability for products.
The main area of research is to find an augmented reality solution for placing 3D furniture in runtime defined real world environments. The solution needs to be mark-less, stable, supportive wide-range of devices, user-friendly, secure and consistent. The current augmented reality solutions does not provide means in which to properly utilize marker-less augmentation. This research based project tries to identify methodologies to provide marker-less augmented reality on a mobile platform and provide a solution where customers can view and customize 3D furniture on their desired location and identify whether the relevant real furniture might actually suit the location.
Provide customers with a mobile application that provide 3D models of furniture which can be placed in their home to identify where it suit them before purchasing actual product.
Provide a means to customize the furniture with the respective customizations provided by the dealer.
Voice recognition feature to allow customer the ease of customizing 3D models placed in the scene.
Provide a suggestions feature that identify the real world location’s main colour and provide furniture suggestion to suit it.
A sound system that can produce high quality directional sounds and simulated sound fields.
Devices like tracked gloves with pushbuttons to enable users to specify their interactions with the virtual objects.
Scan the real world environment and provide data necessary for marker-less AR.
Identification of position and scale of object with respect to the surrounding.
Provide means to add multiple object onto the environment.
Identify and record colour histogram to be used in suggestion for furniture.
Suggestions feature which identify prominent colours and recommend most suitable furniture for the environment.
Provide customizing options via Voice command system.
Provide details in augmented voice format. (Text to Voice)
Provide details of furniture (Prices, Measurements, and any relevant details) next to the respective 3D furniture in the augmented environment.
Provide services that allow filter furniture search based on budget and other space specifications.
Marker-less Augmented Reality relies on tracking features [1] (distinctly identifiable location such as shapes, edges and point) of the area of the real-world retrieved on run time and data from device sensors to identify the position onto which digital images are augmented. Several techniques have been research in Marker-less AR [2], [3], [4] regarding how to effectively map and retrieve features for real-world.
Simultaneous Localization and Mapping (SLAM) is an approach which was originally used in environment mapping for robotics [5], which was latter improved for computer-vision based applications such as AR. since then several methods have been used for achieving SLAM some of which are [2], [3], [4]. Parallel tracking and mapping (PTAM) [2] used a key-frame based Mapping framework which is PC-based, due to this factor PTAM can be used for mobile implementation but with a certain amount of limitations as stated in [3]. Fast-SLAM is another algorithm [4] used to tackle the issue of concurrent mapping and localization. This method has been proven to be operation with small number of features making this a more favorable method to be used in mobile applications.
Another Research area in proposed system is Indoor location tracking. The key part of this system is mapping the indoor environment as it will be the input for all other subpart of the system. Sisi Zlatanovaa, George Sitholeb, Masafumi Nakagawac, Qing Zhud[1] considered existing and emerging problems in indoor mapping. Tough the system based on augmented reality they concluded that in order to transfer more information to user system must use new visualizing techniques.
C. Holzmann and M. Hochgatterer[2] found a way to measure dimensions of objects using mobile camera by determining the distance between the camera positions. They observed that lighting conditions and unintentional device rotations influenced the results. A “baseline” is used to estimate which is based on the accelerometer reading in combination with the gyroscope data and increasing the baseline leads to high accurate results. The device must contain both gyroscope and accelerometer in order to perform.
Another research area in the proposed system is the image processing; which is utilized in the furniture suggestion feature (Explained further in methodology). The color model plays a prominent role during the image processing whereas according to Gonzalez, R.C. and R.E. Woods [6] there are three main models used in modern day computer vision applications. Namely RGB, HIS and CMYK. Each and every model has a different impact on efficiency of image processing. Throughout our project we are using RGB color model which will be an advantage in separating red, green, blue color channels.
During image processing each and every pixel need to be analyzed for its intensity values. Previous research has been conducted by Chhaya S.V., Khera S., Kumar P., [5] revealing how image can be analyzed to capture the intensity values is conducted only for similar single colored image. During this methodology we are supposed to analyze images with different color combinations.
The next research area include voice recognition and augmented audio. The android voice recognition system, according to Yu Zhong, T.V. Raman, Casey Burkhardt, Fadi Biadsy and Jeffrey P. Bigham [6] has an android universal voice control assistance running on devices by synthesizing set of command from on-screen context, and it supports chaining of multiple command in same utterance. It is utilized for commonly used mobile applications .i.e. Sending emails, calling, opening applications, but what our solution utilizes is voice recognition for customization. The voice commands that will be applied in our application to do tasks, within the application’s environment and not in global environment.
Augmented Audio is a wide area to research, in our project we are focusing on how to apply in augmented audio to Imaginary 3-D model that exists in augmented world for android mobile devices. Augmented reality audio can be used in two main ways one is as a part of aural augmentation, and the other as part of User Interface. As an assembly demo [7] the system will give feedback with beeps. According to LookTel [17], the visually impaired better understand the augmented environment which is smart phone applicable. There taking into account these factors and combining in this application we will be using augmented audio as a means to narrate the details of 3D furniture.