<p>Most conventional edge detection methods give the same relevance to the intensity changes (SRIC) in dark (lower intensities), bright (higher intensities) or intermediate (shady intensities) image regions. However, in regions where the grey-levels of pixels are very close, such SRIC concept could have serious consequences. Therefore, there remain miss edge points and spurious responses in the final results. To alleviate this problem, we propose an algorithm which attributes different relevancies to intensity changes (DRIC) in image regions and merges edge results. The edge detection process is achieved in three processing steps. The first step assigns the DRIC in image regions according to an algebraic product method in conjunction with three boosting functions, which we define. The second step determines the gradients magnitude and direction by applying the Sobel operator on any estimated image region. The last step merges the obtained edges at two adjacent smoothing scales. The new proposal is evaluated subjectively and objectively, and has proved to be competitive.</p>