Adaptive Edge Detection Technique Implemented on FPGA
- PDF / 1,917,084 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 105 Downloads / 314 Views
RESEARCH PAPER
Adaptive Edge Detection Technique Implemented on FPGA Saeid Taslimi1 • Rasoul Faraji1 • Ali Aghasi1,2 • Hamid Reza Naji1 Received: 17 February 2019 / Accepted: 12 March 2020 Ó Shiraz University 2020
Abstract This article proposes an adaptive model that improves the edge detection operation in digital images. One of the disadvantages of traditional edge detection operators such as Prewitt, Sobel and Canny is using fixed-size masks which limit the edge detection operation in images with different degrees of the gray levels. In the proposed model, the image histogram is used as a criterion for selecting a suitable mask according to image specifications. In this model, the mask size of the edge detection operator is determined by measuring the contrast of the image to improve the quality of edge detection. For this purpose, a mask with a small size is used for images that have high contrast to increase processing speed, and for the images that do not have high contrast larger masks are used to increase the accuracy of edge detection. The proposed model is implemented utilizing real-time processing and parallelism capabilities of the Xilinx-Virtex6 FPGA. And, in order to reduce the effect of parallelism on increasing the area size of implemented hardware resources, existing resources are reused appropriately. The results show that with the use of the proposed method the edge detection is done in just 2.2 ms and a trade-off between the quality of detected edge and the hardware resources is established. Keywords Image processing Edge detection Real-time processing FPGA Parallelism
1 Introduction Edge detection is one of the most fundamental issues and plays an important role in image processing, which is done in order to extract the salient features of images. Also, it is one of the main tasks of low-level image processing which is directly related to performance, accuracy and runtime. So, with regard to the time-consumption of image processing due to massive data processing, optimum methods in processing and implementation should be utilized to achieve real-time processing (Khan et al. 2015; Li et al. 2015; Peker et al. 2016; Zolfaghari and Yazdi 2014; Khosravi and Momeni 2018; Saryazdi 2016; Roy and Pal 2019). Derivative operators are simple edge detection operators which have different characteristics and detect different
& Rasoul Faraji [email protected] 1
Department of Electrical and Computer Engineering, Graduate University of Advanced Technology, Kerman, Iran
2
Department of Computer Science, Shahid Bahonar University of Kerman, Kerman, Iran
edges depending on the incoming input image. Sobel operator is one of these operators which operates based on differentiation. This operator uses to detect edges. Sobel operator is more applicable than other traditional edge detection operators because of the simplicity and higher precision (Hou et al. 2011). However, it has low accuracy in edge detection because of the use of two fixed masks in both horizontal and vertical directions. To incr
Data Loading...
