International Journal on Document Analysis and Recognition (IJDAR) https://doi.org/10.1007/s10032-020-00357-x ORIGINAL PAPER Automatic room information retrieval and classification from floor plan using linear regression model Hiren K. Mewada 1 · Amit V. Patel 2 · Jitendra Chaudhari 2 · Keyur Mahant 2 · Alpesh Vala 2 Received: 10 December 2019 / Revised: 8 June 2020 / Accepted: 18 July 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020 Abstract The automatic creation of a repository of the building’s floor plan helps a lot to the architects to reuse them. The basic approach is to extract and recognize texts, symbols or graphics to retrieve the information of the floor plan from the images. This paper proposes a floor plan information retrieval algorithm. The proposed algorithm is based on shape extraction and room identification. α-shape is used for finding an accurate shape. From the detected shapes, actual areas of rooms are calculated. Later, a regression model-based binary room classification model is proposed to classify them into room-type, i.e., bedroom, drawing room, kitchen, and non-room-type, i.e., parking porch, bathroom, study room and prayer room. The proposed model is tested on the CVC-FP dataset with an average room detection accuracy of 85.71% and room recognition accuracy of 88%. Keywords Floor plan · Image retrieval · Regression model · α-shape · Document image analysis · Pattern recognition 1 Introduction A floor plan is a graphical representation of top view of a house or a building along with a necessary dimensions. Two- dimensional floor plan evaluation and information retrieval can help in many applications, e.g., to generate 3D model visualization develop virtual-navigation inside the building, count the number of rooms and their area and architectural information recovery. It also provides details of the interior . Architects are required to customize floor plans to meet the client’s requirements. Currently, many architects resort to the tedious process of drawing up plans manually. This cumber- some burden can be significantly reduced with automation, e.g., retrieval of digital plans from a database. This can assure the re-utilization of the existing design for their projects. In addition, previously all floor plans were prepared on the drawing sheet and digitized through photographs. These pho- tographs cannot be read by an application or algorithms, B Hiren K. Mewada hmewada@pmu.edu.sa 1 Electrical Engineering Department, Prince Mohammad Bin Fahd University, Al Khobar, Kingdom of Saudi Arabia 2 CHARUSAT Space Research and Technology Center, Charotar University of Science and Technology, Changa, Gujarat, India thereby rendering them practically useless for further cus- tomization. The algorithms used for automated information retrieval from the floor plans can be divided into two categories: (a) text and label, symbol extraction from the floor plan, i.e., tex- tual processing, and (b) topology connectivity identification and information retrieval, i.e., graphical processing. These two categories belong to image segmentation, layout forma- tion and graph recognition. Identification of labels, symbols and texts from the floor plan was used to get information about the floor plan. This textual information plays a funda- mental role in categorization of the floor plan. Specifically for technical drawing, this approach requires the separation of texts and symbols from the map of the floor plan. However, the occlusion of text with symbols makes this task challeng- ing and in the second alternate of graphical processing, the variety of shapes may create ambiguity in the identification process. Information about the room highly depends on how accurately the shapes are extracted in this second approach. Hence, a robust approach is required to handle these chal- lenges. The proposed graphics approach in the paper uses a robust room’s shape extraction model which distinguishes the rooms efficiently and the regression model is capable of identifying the room-type accurately. The overall contribu- tion of the authors’ is as follows: 123