A Method of Identifying the Abnormalities of Plane with Laser Light Grids System Using Verilog HDL

Transcription

1 A Method of Identifying the Abnormalities of Plane with Laser Light Grids System Using Verilog HDL Rifat Ul Alam. Khan, Sabbir Shirajus. Salekin, Imtiaz Ahmed. Khan, Tabia. Hossain and Iqbalur Rahman. Rokon Abstract Verifying the pattern of the road is one of the biggest challenges for engineers. Using a laser light system to form grid blocks on the road is a very effective way of verifying the pattern and abnormalities of the road. Traditionally, different methods were used to find the abnormalities of road but using a laser light grids block system; it is easier and less complicated. A laser Grid system produces some blocks of square grids on a road by projection light. The image of the square grids projected on the plain road is read by a camera and stored in the ROM in the system. As the person walks by, if there is any abnormalities, or an obstacle on the road, then following deformation is also implicated on the laser grid and the new image formed is again read by the camera, and it is compared with the original image that was previously stored in the ROM. By this process an obstacle ahead, or an abnormalities on the road, can be detected. A pre-alert can be obtained using this system and a person or a driver can be previously alerted about presence of an obstacle or dangerous abnormalities of a road. There are many technology used to find the abnormalities of a road, such as using ultrasonic technology, accelerometer, GPS, smart phone, etc but here we have found a completely different and a new way to find the abnormalities of road. We have used traditional image processing algorithms to develop this system and make it simpler and more users friendly. Keeping hardware implementation in mind, we have designed the entire system using Verilog HDL to ensure flexibility so that our completed and synthesized model can be implemented in devices such as CPLDS, FPGAs and ASICs. Keywords ASICs, CPLDS, FPGA, Laser Light Grids, Abnormalities of Plane, Pattern. Rifat Ul Alam Khan is a final year student of Electrical and Computer (phone: ; rifat.khan@northsouth.edu ). Sabbir Shirajus Salekin is a final year student of Electrical and Computer (phone: ; sabbir.salekin@gmail.com). Imtiaz Ahmed Khan is a graduate from the Electrical and Computer (phone: ; imtiaz2khan92@gmail.com). Tabia Hossain is working as a Laboratory Instructor in Electrical and Computer Engineering Department, North South University, Dhaka , Bangladesh. (phone: ; tabia.hossain@gmail.com). Iqbalur Rahman Rokon is a senior faculty member of Electrical and Computer Engineering Department, North South University, Dhaka , Bangladesh. Former Sr. Engineer, VLSI Chip Research and Development (R&D), Emulex Corporation, California, USA. (phone: ; e- mail: iqbalur.rahman@northsouth.edu). I I. INTRODUCTION N our thesis we have developed a completely new way of finding the abnormalities and obstacles of road using FPGA. This method can be used by cars, motorcycles and most importantly a blind person walking on a street. We used Lumigrids [1] in our thesis to find the obstacles and abnormalities on a road. Lumigrids projects square grid on the road which looks like a net projected on a road with a certain number of rows and columns. The advantage of using this grid system over traditional lights is that, traditional lamp lights casts shadow on both concave and convex obstacle, so it becomes difficult for a computer to read and analyze data from it. A computer won t be able to compare and differentiate a smooth and bumpy road using this traditional lamp. Using grid projection system it is lot easier to identify the nature of the abnormalities, with the squares of the grid bending and changing in an easy way If the road is plain, the shape of the lines in grid won't change and it will show a regular pattern. But in case of abnormalities and presence of obstacle in a road, the shape of the lines will bend and computer will be able to identify the changes. This is done by comparing the current image of the laser grid with the original image of laser grid stored in ROM. When the current image of the laser grid on the road is compared with the stored image in ROM a difference in the grid will be identified and spotted immediately. This is done in real time. The whole grid is divided into six parts Left and Right are the most basic part of division. When there is an obstacle on the left side of the grid, that is, on the left side of the blind person, an electronic beeper beeps on his left ear and if there if there is any obstacle on the right side then the beeper responds on the corresponding ear. 36

2 A. Load II. INTERFACE OF TOP BLOCK DIAGRAM Fig: Block diagram of the overall system It is used to load an image from any external device, camera or storage into the RAM in our system, that is, prepro.v. It is used only once in the entire process just to bring an image into the systems. B. Subtract This module subtracts the image found from the RAM and the default image stored in the ROM. This module actually finds the difference between the images and helps to detect the presence of obstacles or deformation or any change in the expected situation. C. Processing It gets an input of 8 bits, segments the resultant output from subtractor module into 6 equal divisions, constraints a certain % of threshold and in the final stage of the design it gives a 4 bits output. D. Preprocessing It is a RAM we have created to store the image that we get from the camera. The image being processed is stored here pixel by pixel before going through each step in the process. E. Postprocessing It is a ROM we have created to be used in which a fixed image of the perfect plain road is present. III. DESCRIPTION The system is set to work when the enprocessing signal is set to 1 from 0 which triggers the top module to start functioning. We have set apart modules named image.v and dimage.v which would be the primary images (for two individual memory registers). Both images have a predefined dimension of pixels. We have a RAM named as prepro.v, which would call the dimage.v all throughout the processing steps and a ROM named as postpro.v, with the prestored image.v to compare periodically with the RAM. When the top module starts functioning, an image is transferred from image.v to prepro.v with the help of load.v which takes each pixel at every positive clock edge, and transfers it, pixel by pixel, to prepro.v. This process within this module is complete when the signal done corresponds to be 1. Next step is subtraction, which occurs when subtractor.v is enabled. It is enabled when two different images are completely loaded within the two different memory registers. Subtraction module applies the digital numeric value of each pixel from one image to be subtracted from corresponding pixel of another image. In this case, dimage.v is subtracted from image.v, pixel by pixel. The function of subtraction is done within the subtract.v module and the subtracted output image is again overwritten within the prepro.v memory block and thus the signal done implies to be 1. The last and the most important module of our design is the processing.v module, where, initially, the subtracted binary image of the prepro.v module is analyzed and segmented to six equal divisions, as discussed above. We have set some priority in our design so that the most important issues can be resolved first. We have set D1 as the highest priority, i.e. if there is any object or deformation in both D1 and D3, as D1 is closer to the user, it gets the highest priority to get the user informed about obstacle in D1. The priority is set as according: D1>D2>D3. In other words the closer ⅓ of the image gets more priority than the distant ones. Then, according to 6 different segments, 6 different outputs are expected which causes indications for the user to change directions, due to obstruction. If no obstruction is traced, then the output corresponds to nothing. The flagship is finally raised when the signal done marks to be 1 and the whole process is ready again to take a new image into the RAM. Our design is programmed to currently analyze between two images and give outputs corresponding to the different segments within the image. The process currently ends here, but it makes way for an efficient method that we have come up with for matching an image with a stored image in the database. The segmentation of the output subtracted image can be chosen to be in various sizes. If we compare an image with an exactly same image afterwards, each pixels and sizes of each images will match. Since every deformation signifies unique changes within the image, the output subtracted image will never be the same for a different abnormality within the plane, thus ensuring a perfect method of differentiating plane deformations within a set of system. A. Top View Fig: Top view section of the system 37

3 B. RTL Schematic D. Design Hierarchy Fig: Design hierarchy of the whole system IV. TESTING AND EVALUATION Fig: RTL schematic screenshot of the top module (Internal block diagram of the top module) C. Design Flow Fig: A section of the default uniform image Fig: A section of the incoming deformed image The two images used was a 60*90 pixels image, where a portion of each image s screenshot are shown above. The first image is a reference of the default plain road image with symmetrical laser light grids. The other image is a projection of non-uniformed unsymmetrical image (mostly on the right half portion) which refers to some abnormalities that might be for the surface. These two images are hence compared or subtracted with each other to see the 4 bits output. Fig: Design Flow diagram 38

4 V. RESULT Fig: Simulation waveform of top module for deformed image To ensure that our laser light grids system identifying deformation on surface works properly, we have verified the functions by simulating and synthesizing our design using ModelSim-Altera 6.5b and Xilinx 14.7 software respectively. During HDL simulation, the simulator software, verified the functionality and timing of the design. The simulator interpreted the Verilog codes into circuit functionality and displayed logical results of the described HDL to determine the correct circuit operation. Screenshots of a particular image and the waveforms of simulation of the design using that particular image and a default image is shown below. The default image we have used was compared with the incoming image, in this case, we have used a unique deformed image with which our system makes the image undergo the series of processing steps described earlier and gives the desired output, which can be seen in the waveforms. Once the acquisition is over, the done signal, which was 0 until now, changes to 1. VI. DESIGN METHODOLOGY In this section, the design procedure and the architecture of our project has been described. It shows the different stages of the design. The Verilog Codes were at first simulated with ModelSim and then synthesized with Xillinx 14.7 and Quartus II 15.0 device. FPGA technology is chosen because it provides some important advantages over general purpose processors and application specific integrated circuits (ASICs). To simulate the design, the design under test (DUT) and the stimulus provided by the test bench were required. The test bench was an HDL code that allowed providing a documented, repeatable set of stimuli that was portable across different simulators. The test bench used for this design was a more complicated file that included error checking, file input and output and conditional testing. After simulating the design, synthesis was performed. The Quartus II 15.0 Synthesis tool and the Xilinx Synthesis Technology (XST) of Xilinx ISE 14.7 software tool both synthesized the Verilog codes to create Xilinx / Quartus-specific net-list files. VII. ADVANTAGES The main advantage of our project is it is a cheap and user friendly device. Now a days expensive devices like GPS, accelerometer are used to help blind people walking on the street, But the material of our project is cheap and easy to use. There are many technologies to verify and find the pattern of a plane of roads. Various methods can be used to find the plane of a road or to help blind people while walking. This project is a very effective way to find deformation on a road. This overall structure is composed by first image acquisition in the preprocessing. Subtraction module then operates comparing images from preprocessing and postprocessing. The resultant images is overwritten within preprocessing and then finally carried to the processing module which gives a 4 bits signal according to the priority level. Each signal corresponds to unique number of sound beeps given from the headphone. Each step of this structure plays an important role in the whole system and their functions have been studied to best respond to this project aim, which is to create an efficient low cost design to find deformations on road. VIII. CONCLUSION Verilog allows use of RTL description that provides designer advantages while debugging, as the RTL description can be readily edited by the designer & implemented again with small cost of time. The wave form obtained is as it was desired. Therefore we can conclude that the project was successfully accomplished and is ready for further improvements and implementation. Acknowledgment The successful completion of this project would not have been possible without the help of several key individuals. First, we would like to express our gratitude to our supervisor, Iqbalur Rahman Rokon, who met with us monthly throughout the course of the project offering advice and wisdom. Secondly, we would like to thank Imtiaz Ahmed Khan (North South University) for providing us with his valuable advice every time we faced some sort of difficulty while programming the design, and Dr. Lamia Iftekhar (North South University) for her continued appreciation and support. REFERENCES [1] Simpson hill. (June 20, 2014). 5 Amazing Gadgets that are Helping Blind to see. Retrieved from [2] Red Dot Award. (2012). Gizmag. Lumigrids- The LED Projector that keeps cyclist out of Pit holes. Retrieved from [3] Khan, I. A., Chowdhury, M. A., & Rokon, I. R. (2015). A distinct approach to personal identification based on vein pattern recognition using Verilog HDL. International Journal of Engineering Research and Technology Rifat Ul Alam Khan is a final year student of Electrical and Computer Engineering Department, North South University, Dhaka , Bangladesh. His major is in VLSI. He undergone through various VLSI based projects enrolling through his undergraduate curriculum and he is keen to learn more regarding such topics. Sabbir Shirajus Salekin is a final year student of Electrical and Computer Engineering Department, North South University, Dhaka , Bangladesh. His major is in VLSI. Creative thinking and problem solving in Engineering is his passion. 39

5 Imtiaz Ahmed Khan is a graduate of Electrical and Computer Engineering Department, North South University, Dhaka , Bangladesh. His major is in VLSI. He holds up an extraordinary knowledge in terms of Verilog HDL language and his advice holds up an extra privilege for this project. Tabia Hossain received BSc degree in Electrical and Electronics Engineering Department from North South University in Currently she is doing MS in EEE at NSU and also working as a Laboratory Instructor in Electrical and Computer Engineering Department. Iqbalur Rahman Rokon received his BSc degree in Electrical and Electronics Engineering from Bangladesh Institute of Technology, Rajshahi, Bangladesh in 1991 and MS degree in Electrical and Computer Engineering from California University, Northridge, USA in During his MS study at USA, he also worked as System Engineer at National Telecom, Santa Ana, California, USA from 1994 to He then started his career at High-Tech chip companies in USA and served as Design Engineer. Chip (ASIC/FPGA/CPLD) Development, R&D, Emulex Corporation, Costa Mesa, USA from 1997 to He then joined North South University, Dhaka and has been teaching there as a Faculty Member of Electrical and Computer Engineering Department since He also served as a Proctor of the university from 2007 to In addition to his current teaching profession in NSU, Iqbalur Rahman. Rokon also worked with hightech industry and served as consultant of ASIC/FPGA Development at Power IC Limited: A chip design company at Dhaka, Bangladesh 40