ADVANCED MATERIALS & PROCESSES | APRIL 2024 38 iTSSe TSS iTSSe TSS FEATURE 9 Digitized shapes were then imported into SolidWorks and the final artwork color palette and design were predicted. CAD designs were laser machined (KJ Laser Micromachining, ON, Canada) into 1/16-in. thick stainless steel sheets to minimize mask deformation and powder adhesion to the mask surface. COLD SPRAY PROCESS The PCS100 cold spray system (Plasma Giken, Japan) was used to produce all coatings. All final depositions were completed at Polycontrols (Quebec, Canada) facilities equipped to manage and spray large components. Table 1 shows the spray parameters used for each powder material. The gun traverse speed was selected based on the intended local coating thickness for each region of the art piece. Particle velocity, shown in Table 1, was measured using a Coldspray Meter (Tecnar, QC, Canada). RESULTS The aluminum canvas plate was first coated with pure aluminum powder to provide an irregular roughened artistic background reminiscent of the photographed fossil. Use of the irregular pure aluminum powder led to complete coating delamination during the deposition of subsequent layers. This issue was resolved by recoating a new aluminum 6061 plate (32 x 26 x 1 in.) was chosen as the substrate/ canvas. Prior to spraying, the plate was grit-blasted and cleaned with ethanol and compressed air. DIGITIZATION AND MASK CREATION Digitization of the fossil image was completed through color palette and shape recognition using the 2D photograph shown in Fig. 1. First, the 8-bit RGB (red, green, and blue) photograph was converted to an 8-bit grayscale image using Fiji ImageJ open-source software. Next, image segmentation based on gray level information was conducted during which shape and color detection were completed by grouping pixels of similar metrics to a sample group (Fig. 3). In this step, the luminescence of each pixel was compared to the user set threshold. Groups falling within the analyzed threshold category were representative of a region, scene, or object recognizable to the eye. If the region of interest was successfully highlighted through the general threshold method, a nonlinear median filtering algorithm was used to smooth, sharpen, and suppress unnecessary noise while preserving the region’s original information. Otherwise, if the general thresholding was unsuccessful at properly providing sufficient region delineation and definition, local thresholding was used along with the implemented plugins in ImageJ[4-7]. Fig. 2 — Top row: Powder SEM images with 25 µm scale marker at bottom of figure. Bottom row: Powder photos. From left, spherical Ti64, agglomerated nickel, spherical nickel, irregular zinc, irregular steel, spherical aluminum, nickel blend, and irregular aluminum. Fig. 3 — From left, global/general threshold selection process using level 128; binarization from selected threshold values; local threshold applied on a distinct feature; final CAD predictive color palette and design; and laser machined mask.
RkJQdWJsaXNoZXIy MTYyMzk3NQ==