About 10% of the population have some form of color vision deficiency. One of the most sever deficiencies is dichromacy. Dichromacy impairs color vision and impoverishes the discrimination of surface colors in natural scenes. Computational estimates based on hyperspectral imaging data from natural scenes suggest that dichromats can discriminate only about 7% of the number of colors discriminated by normal observers on natural scenes. These estimates, however, assume that the colors are equally frequent. Yet, pairs of color confused by dichromats may be rare and thus have small impact on the overall perceived chromatic diversity. By using an experimental setup that allows visual comparation between different spectra selected form hyperspectral images of natural scenes, it was estimated that the number of pairs that dichromats could discriminate was almost 70% of those discriminated by normal observers, a fraction much higher than anticipated from estimates of the number of discernible colors on natural scenes. Therefore, it may be rare for a dichromat to encounter two objects of different colors that he confounds. Thus, chromatic filters for color vision deficiencies intended to improve all colors in general may constitute low practical value. On this work it is proposed a method to compute filters specialized for a specific color-detection task, by taking into account the user’s color vision type, the local illuminant, and the reflectance spectra of the objects intended to be distinguished during that task. This method was applied on a case of a medical practitioner with protanopia to idealize a filter to improve detection of erythema on the skin of its patients. The filter improved the mean color difference between erythema and normal skin by 44%.