2 Observations

2.1 Target selection

Our target list was built using the catalog of compact radio sources in Cygnus X of Wendker et al. (1991). Based on the spectral index measured between 408 and 4800 MHz, that paper classifies the sources as thermal (HII regions) and non-thermal (supernova remnants). Only the first class was considered for the present study. Furthermore, we retained only those sources whose listed size was compact enough to be completely covered in the field of view of the detector used (see Sect. 2.1), thus rejecting HII regions whose listed size exceeded 5$^\prime$. Some of the sources meeting our two selection criteria have been intensively studied by other authors, also in the near-infrared, and therefore we excluded them from our target list as well. These sources are S106, DR21, and DR18. The targets in our final sample are individually discussed in Sect. 3.3, and their coordinates and main features are given in Table 1 at the end of this paper. For the sake of homogeneity we use the object denominations of Wendker et al. (1991), although some sources, noted in Table 1, also have more commonly used DR (Downes & Rinehart 1968) denominations.

2.2 Observations and data reduction

The observations presented in this paper were carried out on the nights of 6 and 7 August 1998 using CAIN, the near-infrared array camera at the 1.5 m Carlos Sánchez Telescope in the Observatorio del Teide (Canary Islands, Spain). This camera has a Rockwell $256 \times 256$ array with a pixel scale of 1 $^{\prime\prime}$00 per pixel at the Kband (the scale depends slightly on the filter used), yielding a field of view of $4'2 \times 4'2$ per frame. Each observation of a region in a given filter consisted of 9 sequences of 5 individual frames each, obtained on a dither grid of $3 \times 3$ points in a square pattern of $20\hbox{$^{\prime\prime}$ }\times 20\hbox{$^{\prime\prime}$ }$. The 5 individual frames at each position, each of 6 s of exposure time, were saved individually. This was preferred to stacking of the 5 exposures in a single frame, due to the appreciable drift in the positioning of the telescope during the 30 s spent at each position: blind tracking was used, as the frequency of offsets to complete the drift pattern would otherwise have imposed a large overhead in the execution time if using the autoguider. Shift-and-add at the reduction stage allowed us to compensate for the drift, as well as to remove cosmic rays and detector bad pixels in the stacked images. Sky images were obtained separately, as the crowdedness of our target fields and the abundance of extended nebulosity did not allow us to construct acceptable sky background images by median-filtering of the jittered exposures on the target fields. The observing strategy employed was as follows: first, all the HII regions were imaged in the Kfilter, in a sequence consisting of one observation of one field, a nearby sky region, another field, the sky region again, and so on until observing all the fields in our list. Then, the sequence was repeated in the H filter, and finally in the J filter. Photometric calibration was obtained from short exposures of each field in each filter on both nights when the Cygnus X region was near the meridian, alternating the exposures of the fields with observations of the infrared standard star FS 29 (Casali & Hawarden 1998).

The observations were reduced using standard infrared imaging data reduction procedures, implemented by means of dedicated IRAF scripts. The small variations in the scale of the images obtained through the different filters were corrected for by enlarging the images in the J and H filter so as to match the scale of the K image, using a flux-conserving pixel interpolation schema. Point sources were detected by adding the frames in J, H, and K, and running DAOPHOT (Stetson 1987) on the resulting image. The photometry was obtained as follows: in a first step, aperture photometry with a large aperture was performed on the images of the standard star and of suitable bright and isolated stars in the short exposures of the fields imaged at the same time as the standard. This allowed us to set up a network of bright secondary standards in each field and filter. Then, aperture photometry with a small aperture (3 pixels in radius), adequate for our rather crowded fields, was performed on the deeper images. Finally, magnitudes for all the stars detected in each field were determined taking those of the secondary standards as a reference. We also experimented with results obtained by performing point-spread function (PSF) photometry on the deeper images. However, we found that individual zeropoints obtained from the comparison between PSF photometry of each secondary standard in the deep exposures and aperture photometry in the calibration exposures yielded a considerably larger scatter than when comparing the latter with small-aperture photometry in the deep exposures. The complete list of positions and magnitudes of stars in each region is available electronically only as Table 2 through the Centre de Données Stellaires (Strasbourg, France).

Figure 1: Images, color-magnitude, and color-color diagrams of each region (the full sample can be found in the electronic version of this paper only). Selected samples are shown in the printed copy, where J (left), H (center) and K (right) images are presented for ECX6-24 and ECX6-38. The images display an area of $4'2 \times 4'2$ around the position of the emission peak at 4.8 GHz, as listed in Table 1. North is up, East to the left. The boundaries that we propose for the ionizing stellar aggregate are outlined in the K-band image when such an aggregate can be discerned. No color-color and color-magnitude diagrams are presented for ECX6-7a, due to the presence of the heavily saturating star BC Cyg. In the color-magnitude and color-color diagrams each dot represents a star in the field, and filled circles are stars within the adopted boundaries of the aggregate. The straight lines in the diagrams represent the direction and amount of the shift in the position of a star caused by a visual extinction of AV = 10 mag, according to Rieke & Lebofsky (1985). The curves in the color-color diagrams mark the locus of unreddened main sequence stars (bottom branch) and giants (upper branch) using intrinsic colors from Bessell & Brett (1988). This locus is not marked in the color-magnitude diagrams, as the distances are unknown, but the early-type type stars of interest in the present work would lie if unreddened along a nearly vertical line at $H-K \simeq -0.04$. To minimize the scatter due to faint stars near the detection limits and to spurious detections, only stars with errors $(\Delta H, \Delta K) < 0.2$ mag are represented in the color-magnitude diagrams, and with $(\Delta J, \Delta H, \Delta K) < 0.2$ mag in the color-color diagrams.