The integrity of the reconstructed human epidermis generated in vitro could be assessed using histological analyses combined with immunohistochemical staining of keratinocyte differentiation markers. Computer-based analysis of scanned tissue saves the expert time and may improve the accuracy of quantification by eliminating interrater reliability issues. However, technical differences during the preparation and capture of stained images and the presence of multiple artifacts may influence the outcome of computational methods. Using a dataset with 598 unannotated images showing cross-sections of in vitro reconstructed human epidermis stained with DAB-based immunohistochemistry reaction to visualize 4 different keratinocyte differentiation marker proteins (filaggrin, keratin 10, Ki67, HSPA2) and counterstained with hematoxylin, we developed an unsupervised method for the detection and quantification of immunohistochemical staining. The proposed pipeline includes the following steps: (i) color normalization to reduce the variability of pixel intensity values in different samples; (ii) color deconvolution to acquire color channels of the stains used; (iii) morphological operations to find the background area of the image; (iv) automatic image rotation; and (v) finding markers of human epidermal differentiation with clustering. Also, we created a method to exclude images without DAB-stained areas. The most effective combination of methods includes: (i) Reinhard's normalization; (ii) Ruifrok and Johnston color deconvolution method; (iii) proposed image rotation method based on boundary distribution of image intensity; (iv) k-means clustering using DAB stain intensity. These results should enhance the performance of quantitative analysis of protein markers in reconstructed human epidermis samples and enable comparison of their spatial distribution between different experimental conditions.