A novel edge detection method based on image energy and skewness with application to intramuscular fat recognition

Effective edge detection algorithms are important in image segmentation processes, because they increase the success in identifying objects and save the computational time in the further processing steps. One drawback of the common gradient edge detectors is related to the smoothing step of the original image. These detectors whether ignore this step, producing noise- sensitive detection results, such the detection with Roberts, Prewitt, and Sobel detectors, or apply a Gaussian smoothing filter, at which the detection results are dependent on the chosen filter size, such the detection with Canny detector. In this paper, a novel edge detection method is presented based on image energy and skewness as two smoothed versions of the original image. The method has been tested for various types of real world images. The experimental results gave at least 6.451% increment in the signal to noise ratio, and 1.667% reduction in the root mean square error, in comparison to the results of Canny edge detector. Consequently, the method had been applied efficiently to detect the intramuscular fat contents in a meat slice image, which is a quite complex problem due to the difficulty of the meat-fat structures interference.