Enhancement of Gradient for Touching and Overlapping Objects

Normally, gradient image is sometime poor especially for touching and overlapping objects. The proposed method manipulates the use of HSV channels in order to enhance the gradient image obtained for objects with complex background. Only S and V channels are used in this technique. The proposed gradient enhancement approaches were tested on HSV channel images. The Standard Sobel and Laplace of Gaussian (LoG) technique was used to compare the results and performance. The Mean Squared Error (MSE) and Peak signal-to-noise ratio (PSNR) values were calculated as a performance matrix for the implementation. The results show the improvement of the gradient image if compared to existence approach.