An image forming method and apparatus capable of implementing accurately and constantly an improved conversion of the amount of toner adhesion over the entire range of the amount of adhesion. The method includes the steps of computing a normalization value as a relative output ratio of the regular reflection output to a background regular reflection component from the surface extracted from the regular reflection light, in which the regular reflection output is obtained by detecting a plurality of gradation toner patterns with a sensor configured to simultaneously detect regular reflection light and diffuse reflection light; obtaining a diffuse reflection output conversion factor by either (1) subtracting the normalization value multiplied by the diffuse reflection output voltage generated by the surface from the diffuse reflection output voltage, or (2) subtracting the normalization value multiplied by a diffuse reflection output voltage increment, which is computed as the difference between the diffuse reflection output voltage and another diffuse reflection output voltage obtained when a light emitting device is turned off, from the diffuse reflection output voltage increment; and subjecting the relation between the diffuse reflection output conversion factor and the amount of adhesion to a polynomial approximation.