A new class of nonincreasing filters for morphological image processing is presented. In fact, the morphological filtering plays a fundamental role in extracting the markers of the images to apply the traditional watershed-plus-marker approach to segment the images. Here, the geodesic transformation and morphological contrast operators (the morphological gradients and the tophat transformations) are also the bases to apply this segmentation approach. In this work, we are interested in these morphological contrast operators and in a class of operators called toggle mappings (in mathematical morphology). These results enable us to propose nonincreasing filters, based on the notion of morphological gradients. Two approaches are used to apply these propositions. First, we study the operators by changing the size of the structuring element and a given ? parameter to obtain toggle mappings. Next, we iterate these operators until idempotence is reached as the reconstruction transformations occur. The filters have interesting properties and give essential contrast to the images. Using these filters, it is possible to modify the minima (maxima) of images and to obtain a good segmentation when the traditional transformation, the watershed, is applied. Finally, the new operators are compared with other transformations to show their relative performance.